--- title: "What Do I Need Hebby For? Use Cases | Hebby Memories" description: "What do I need Hebby for? Examples with and without memory for creators, developers, trades, authors, researchers, job seekers, business and more." canonical: "https://www.hebby-memories.com/en/use-cases/" language: en last_updated: 2026-11-07 image: "https://www.hebby-memories.com/assets/og-bild.jpg" markdown_versions: ["https://www.hebby-memories.com/wofuer.md", "https://www.hebby-memories.com/en/use-cases.md"] publisher: "Hebby Memories, https://www.hebby-memories.com (llms.txt: https://www.hebby-memories.com/llms.txt)" --- Use cases # What do I need Hebby for? Long story short: Hebby makes your AI smarter, more efficient and more pleasant to work with. Bottom line: you get results much faster and boost your productivity. ## What do you use AI for? Then we'll show you with an example why, and how. Click whatever fits you ↓ **All you need:** you use AI and have at least the smallest ChatGPT or Claude plan. Hebby learns from everything you discuss and work out with your AI. The more you work with it, the more it knows. ### But doesn't my AI already come with built-in memory? Yes. ChatGPT and Claude remember notes about you, and agents like Claude Code or Codex read context files such as CLAUDE.md or AGENTS.md at the start of every session. Hebby builds on exactly this principle: memories as text your AI understands directly. We are convinced that this is the only way a memory fits seamlessly into your AI. But Hebby takes it much further, because built-in memory has clear limits: - **Fills up fast.** If you work a lot with your AI, you fill it up soon, with big projects and complex topics often after a short time. Then old memories have to make room for new ones, and much is never kept at all. - **Slows your AI down.** Built-in memory is usually sent in full, whether it fits the question or not. The fuller it gets, the more of your context window it takes up, and your AI becomes dumber, slower and more expensive. - **Separate notes, no network.** Who works on which project, which decision followed from what, what belongs together: built-in memory doesn't know these connections. - **Each on its own.** What ChatGPT knows about you, your Codex or Claude doesn't. Every app and every agent has its own memory. Hebby always shows your AI exactly what matters right now, concise and in a form language models understand particularly well. On top of that come processes from memory psychology: Hebby links what is needed together, sorts overnight and lets go of what is outdated. One memory for all your agents and devices, built for the pace at which AI evolves. Developers & freelancers ## Every project. Every client. Every rule. Hebby knows all your projects and, if you like, the clients behind them: their wishes, what's planned and what's already built. Your agent connects the dots, makes better suggestions and needs less time and fewer tokens for your code. - Up to speed from the first prompt of every session: your agent doesn't have to read through the whole codebase at great token cost. Hebby brings exactly what the task needs, and what is connected to it. - Fewer follow-up questions, fewer old mistakes, smarter suggestions. And project-specific rules apply automatically. - Your client doesn't want their whole codebase going through AI servers? Hebby remembers the state, purpose and open tasks of every file. You still get architecture advice without uploading entire files. - Each agent sees only the project it is working on right now. - Let finished projects rest: while you work on other things, your agent doesn't see them. Need to pick one up again? Switch it back on and your agent knows everything, even after a long time, as if it were yesterday. - Back on a project after months? Your AI briefly explains your own architecture and why you built it that way. It knows where everything is, what works and who did what, without reading the code first. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **Claude Code**Built-in** › You: **Client Weber wants a PDF export for the monthly reports in the dashboard. Build it.** ◇ Built-in memory: 10 entries sent along › AI: Weber has a dashboard for a web shop, and your stack is TypeScript. For the PDF export I'll use jsPDF with a template like the one for Brandt and pull the reports from /v1/reports. I don't know where the export button sits in the Weber dashboard. Can you show me the file? I'll ask later whether a logo belongs in the header. ✗ Takes jsPDF from Brandt and the old route /v1/reports. Both are wrong for Weber. › You: **Client Weber wants a PDF export for the monthly reports in the dashboard. Build it.** CLAUDE.md · always sent in full - - Stack for client projects: TypeScript, Tailwind, deploy via Vercel - - Client Weber: dashboard for the web shop, branches named feature/ - - Tests with npm run test, lint and tsc --noEmit before every commit - - Client Brandt (customer portal): PDF invoices with jsPDF, template in src/pdf/template.ts - - Client Brandt: staging sign-off by Mr. Lindner, no deploys on Fridays - - Client Hartmann (warehouse system): Express backend, table export as XLSX with exceljs in src/export/ - - Weber's backend API is still called /v1/reports (as of spring, migration planned) - - Old project client Roth (Shopify theme): use Node 14, don't touch - - Commit messages in English, Conventional Commits - - Keep answers short, no long explanations helps a littlesent for nothing With Hebby **Claude Code**With Hebby** › You: **Client Weber wants a PDF export for the monthly reports in the dashboard. Build it.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: I'll build the PDF export as a new exporter. Your architecture note tells me where everything sits, so I don't need to read the rest of the codebase. 1. New file lib/export/pdf.ts, registered via registerExporter in index.ts, next to csv.ts. In ReportToolbar.tsx the "Export" button gets a "PDF" entry. 2. You decided not to add a new library without asking, because Weber has every dependency checked beforehand. So I'm using pdf-lib, which lib/invoice/pdf.ts already uses for invoices. 3. The reports API returns at most 500 rows per request. I load page by page and assemble the PDF at the end. 4. The header carries assets/weber-logo.svg instead of a line of text, as Ms Seidel asked. Plus a test under npm run test:export. Shall I prepare the pull request for staging right away? Ms Seidel signs off changes there. ✓ A plan from your own architecture note, without reading the codebase. With the client's rules. › You: **Client Weber wants a PDF export for the monthly reports in the dashboard. Build it.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationProject Weber (dashboard for the web shop): Next.js with TypeScript, export module in lib/export, so far CSV only - informationpdf-lib is already in the project, lib/invoice/pdf.ts uses it for invoices - decisionNo new library without asking Weber first, rather than adding as needed, because Weber has every new dependency checked legally (takes up to 2 weeks) - behaviorThe dashboard's reports API (/api/reports) fails with more than 500 rows per request, so load page by page - eventStefan wrote down the architecture of the Weber dashboard on 18 September, it is in the repo: docs/ARCHITEKTUR.md Do I need the architecture note? Yes, it says where exporters are registered and where the button sits. ARCHITEKTUR.md ARCHITECTURE Weber dashboard As of 18 Sep, Stefan K. app/reports/ report pages components/ReportToolbar.tsx "Export" button lib/export/ index.ts registerExporter(name, fn) csv.ts only exporter so far types.ts Exporter = (rows) => Blob lib/invoice/pdf.ts uses pdf-lib api/reports max. 500 rows per request assets/weber-logo.svg Tests: npm run test:export - decisionWeber logo (assets/weber-logo.svg) in the header of every client PDF, not a text heading, because Ms Seidel (Weber) asked for it on reports to her customers - informationChanges for Weber go to staging first, sign-off by Ms Seidel, never straight to production - informationWeber's requirements document is on the Drive: Weber/Requirements_2026.pdf Do I need the requirements document? No, the PDF request is clear, and the technical side is in the architecture note. Requirements_2026.pdf stays closed - eventCSV export showed broken umlauts in Excel, fixed on 2 March with UTF-8 including BOM usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Developers%20%26%20freelancers) Trades & small businesses ## Your agent knows every job, every site, every agreement. Surveys, customer requests, delivery dates, your rules: whatever your team discusses with the AI, Hebby remembers. Your agent writes quotes that already take everything into account and answers customer questions without you digging through folders. - Quotes in minutes, with survey, framework prices and dates. - Your rules apply automatically, such as when outdoor work is possible. - Knows who was on site and where plans and notes are stored, and opens them only when they are needed. - Customer on the phone? One look at your phone and you know where things stand. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **Lindner Logistics wants a quote for their warehouse in Duisburg. Draft one for me.** ◇ Built-in memory: 10 entries sent along › AI: Sure, I'll draft a quote for Lindner Logistics in Duisburg. As with Ruhrkontor, I'm using €4.80 per metre of forklift lane and addressing it to Mr. Petersen. I don't know the size or scope: how big is the warehouse, and what needs marking? Or do you mean the Rheinhafen warehouse you already quoted in Duisburg? ✗ Takes Ruhrkontor's price and contact. Doesn't know Lindner. › You: **Lindner Logistics wants a quote for their warehouse in Duisburg. Draft one for me.** Saved memories · always sent in full - Runs a road marking business in Essen. - Has two employees, Mike and Tom. - Prepares a quote for Ruhrkontor for 25,000 m² of hall marking, contact Mr. Petersen. - Calculates €4.80 per metre of forklift lane for Ruhrkontor, 3% discount. - Quote for Kowalski Freight in Gelsenkirchen: Ms. Weber wants construction phases listed separately. - Prepared a quote for a warehouse of Rheinhafen Freight in Duisburg, 2,800 m². - Invoice for the car park of Autohaus Brenner in Mülheim sent in October, closed. - Orders marking paint from Lackhandel Schulte, 2-day delivery. - Stadtwerke Bochum job: car park marking finished in September, invoice paid. - Prefers short answers without filler. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **Lindner Logistics wants a quote for their warehouse in Duisburg. Draft one for me.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: Here is the draft for Lindner Logistics, Duisburg warehouse. It is based on Mike and Tom's survey from 12 February. 1. Halls 1 to 3: 541 m of yellow forklift lanes, 12 cm, plus 94 m of walkways in hall 1 and 38 m² of restricted areas in hall 2. The halls are covered, so we can start as soon as the paint arrives on 6 March. 2. Floor repair at gate 4 (approx. 6 m²) as a separate item. Mike and Tom flagged it. 3. I'm scheduling hall 2 for a weekend, because forklifts run there until 10 pm. 4. Outside: the 18 covered bays at the ramp go in now as well. The other 46 bays and 22 arrows don't: you decided to mark only covered areas in winter, because the paint won't dry otherwise. I'm planning them for May and noting that in the quote as a second phase. All prices include the 6 % framework discount. Shall I draft the cover letter to Ms Brandt too? ✓ A finished quote instead of five questions. With survey, dates and your rules. › You: **Lindner Logistics wants a quote for their warehouse in Duisburg. Draft one for me.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationLindner Logistics' Duisburg site: warehouse with 3 halls, 4,390 m², large car park, many forklift lanes in the halls, part of the outdoor area covered - decisionIn winter, mark only covered areas, not outdoors, because the paint doesn't dry in cold and wet weather - eventMike and Tom on site at the Duisburg warehouse, survey is on the server: Clients/Lindner/Duisburg/Survey_Warehouse.pdf Do I need the survey? Yes, the quantities are only in there. Survey_Warehouse.pdf SURVEY Lindner Logistics, Duisburg site Site visit 12 Feb, M. Krause / T. Yilmaz Hall 1 Forklift lanes yellow, 12 cm 186 m Walkways between racks 94 m Hall 2 Forklift lanes yellow, 12 cm 212 m Hatched restricted areas 38 m² Hall 3 Forklift lanes yellow, 12 cm 143 m Outside Parking bays total 64 of which covered (ramp) 18 Road arrows 22 pcs Note: floor in hall 2 at gate 4 damaged, repair first (approx. 6 m²) - eventLane marking paint for Lindner ordered, delivery to the warehouse on 6 March - informationLindner Logistics has a framework contract: 6 % off all list prices - decisionMark hall 2 only at weekends, not on weekdays, because forklifts run there in two shifts until 10 pm - informationNotes from the first meeting with Ms Brandt are on the server: Clients/Lindner/First_meeting.docx Do I need the meeting notes? No, I already have the key facts. First_meeting.docx stays closed - informationMs Brandt wants quotes as a PDF by email, copy to purchasing - informationLindner plans a second warehouse in Moers for 2027 usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Trades%20%26%20small%20businesses) Business & workaholics ## Your agent knows who works with you on what, and when. Connect your agent to email and calendar, and Hebby remembers contacts, project status and agreements without you maintaining anything. Jump between topics just in time, and your agent jumps right along. - Meeting notes in seconds, with everything that belongs to the contact. - Hebby picks up changed project status on its own. - Your agent suggests synergies that get lost in day-to-day work. - From your phone between two meetings to your laptop at the office. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **Prepare my meeting with Ms Krüger.** ◇ Built-in memory: 9 entries sent along › AI: Sure. Frau Krüger from Nordlicht, I know her from your customers in northern Germany. I'll assume an offer like Hansa Pack's: flat 5% discount, 14 days payment terms, a price around €0.58 per piece like the roastery. I don't know Frau Krüger's quantity or what she wants. What is the meeting about? Then I'll adjust the numbers. ✗ Transfers discount, payment terms and price from Hansa Pack and the roastery to Frau Krüger. › You: **Prepare my meeting with Ms Krüger.** Saved memories · always sent in full - Heads sales at a packaging manufacturer. - Has customers in northern Germany, including Nordlicht GmbH. - Prepares an offer for Hansa Pack for 25,000 shipping boxes, contact Herr Petersen. - Hansa Pack gets a flat 5% discount and 14 days payment terms. - Prepared an offer for the Bremer Kaffeerösterei in spring, contact Frau Lindemann, €0.58 per piece. Order completed. - Had a meeting in September with Herr Brandt of Elbe Logistik about delivery times. - Used to want offers as Word files instead of PDF. - Prefers short answers with bullet points. - Is preparing a trade fair in Cologne. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **Prepare my meeting with Ms Krüger.** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: Your meeting with Ms Krüger (Nordlicht GmbH) is tomorrow at 10 am by video, as her invitation in your inbox says. Where things stand from your call on 22 October: 1. She wants 40,000 folding boxes for 2027 (2026: 36,000, over 11 % more). At your €0.62 that's €24,800. The competitor offers €0.60, a little over 3 % less. 2. You decided against a flat discount, because she reads discounts as a sign that the price was too high before. So my suggestion: a volume tier at €0.60 from 40,000 pieces. That's €24,000, she saves €800. 3. You can afford that: Kartonwerk Lumen cuts its price from €0.41 to €0.39 on 1 January, so your margin stays at €0.21 per piece. 4. Add her wish for fixed deliveries each quarter. For the opening, her love of sailing works: the Baltic Sea. Shall I prepare the quote as a PDF, with a copy to Mr Thiel? ✓ Meeting prep with figures, strategy and an opener. From emails and your agreements. › You: **Prepare my meeting with Ms Krüger.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationMs Krüger: head of purchasing at Nordlicht GmbH, client since 2023, folding boxes for shipping, bought in 2026: 36,000 pieces - eventCall with Ms Krüger on 22 October about the 2027 quote, note is at Documents/Clients/Nordlicht/Call_note.md Do I need the call note? Yes, quantity, prices and her wishes are only in there. Call_note.md NOTE Call with Ms Krüger, Nordlicht 22 Oct, 15 min. Topic Quote folding boxes 2027 Quantity 40,000 pcs (2026: 36,000) Our price €0.62 per pc = €24,800 Competitor €0.60 per pc (about 3 % less) Payment 30 days, unchanged Wish fixed deliveries each quarter Meeting Fri 6 Nov, 10 am, video Open wants the figures in writing - decisionLower the price for Nordlicht through a volume tier, not a flat discount, because Ms Krüger reads discounts as a sign that the price was too high before - informationKartonwerk Lumen cuts the purchase price of folding boxes from €0.41 to €0.39 per piece on 1 January - informationMs Krüger wants quotes as a PDF, copy to Mr Thiel in controlling - informationThe 2023 framework contract with Nordlicht is at Documents/Clients/Nordlicht/Framework_contract.pdf Do I need the framework contract? No, the terms are in the call note, and this is about 2027. Framework_contract.pdf stays closed - informationMs Krüger likes sailing, small-talk topic: the Baltic Sea - informationNordlicht moves into a new shipping warehouse in Rostock in summer 2027 usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Business%20%26%20workaholics) Creators & influencers ## Your agent knows every post, every video, every hit. YouTube, Instagram or TikTok: Hebby remembers your content, what it was about, why you made it, what worked and where it lives. Pitch a new idea and your agent instantly has the right context at hand, by association. - From the first prompt of every new session, your agent knows your channel. No costly re-reading of all your videos: like a flash of memory, Hebby brings what fits the idea. - New ideas build on your channel instead of repeating it. - Openings, titles, captions and thumbnails in your style, without explaining your rules every time. - One idea, three platforms: your agent knows what takes off on YouTube, Instagram and TikTok. - Fewer skills and prompt templates: your workflow lives in memory. - Script, edit, posting plan, community replies: every agent knows what the others are doing. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **New idea: a video about travelling cheaply in Japan.** ◇ Built-in memory: 10 entries sent along › AI: Budget travel in Japan fits your channel well. As a title I'd suggest "Japan on €25 a Day: Is It Possible?", like your Bali video. Add a Reel and a short mention of Packwerk, which you agreed with Ms. Lorenz. I don't know what your audience wants, though. Are there questions from the comments I should pick up? ✗ Reuses the title of the weak Bali video. And brings Packwerk back. › You: **New idea: a video about travelling cheaply in Japan.** Saved memories · always sent in full - Runs a travel channel on YouTube, focused on backpacking. - Also publishes Reels on Instagram. - Published the video "Bali on €25 a Day" in June, performed poorly. - Partnership with luggage brand Packwerk for two videos, contact Ms. Lorenz, invoice sent. - Plans "Lisbon in 48 Hours", script due 20 Nov. - Posts Reels on Mondays and Thursdays. - Edits with DaVinci Resolve. - Always adds chapter markers in video descriptions. - Hostel series "5 Hostels Tested" discontinued after episode 3. - Prefers a casual tone, likes to use informal address. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **New idea: a video about travelling cheaply in Japan.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: Make it the sequel to your Thailand hit: "Japan on €50 a Day: Is It Possible?" 1. The demand is there: 118 of the 310 requests under the Thailand video named Japan, far ahead of Vietnam with 74. 2. The daily budget goes into the first seconds, because your viewers drop off during long intros. Then you answer the three top questions: Tokyo, capsule hotel or hostel, rail pass. 3. Add a reel under 30 seconds with your 3 best money-saving tricks. The Bangkok reel brought 2,100 subscribers. You decided to always put an amount in the title because those get clicked twice as often. That's why the 50 is in the title. Shall I prepare two thumbnail versions for your poll? ✓ A video that builds on your hit, with numbers from your own comments. › You: **New idea: a video about travelling cheaply in Japan.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationTop video "Thailand on €30 a Day" (May): 150,000 views, twice the channel average - behaviorViewers drop off when the first 20 seconds are just small talk; videos that show the daily budget right away are watched for longer - decisionAlways put a concrete amount in the title, rather than no number, because titles with an amount got clicked twice as often on this channel - eventReel "3 Money-Saving Tricks in Bangkok" (May): 12,000 views and 2,100 new subscribers, the link to the long video was in the profile - eventComments under the Thailand video analysed, result is on Drive: Channel/Analysis/Comments_Thailand.txt Do I need the analysis? Yes, the exact numbers and questions are only in there. Comments_Thailand.txt COMMENTS Thailand video, as of 14 Sep Total: 1,240, of which requests: 310 Requested next destination Japan 118 Vietnam 74 Portugal 41 Other 77 Top Japan questions 1. How expensive is Tokyo really? 2. Capsule hotel or hostel? 3. Is the rail pass worth it? - informationAudience: backpackers between 20 and 30, watching almost only on their phones - informationScript of the Thailand video is on Drive: Channel/Scripts/Thailand_30_Euro.docx Do I need the script? No, I already know the structure. Thailand_30_Euro.docx stays closed - decisionTest thumbnails in two versions in a community poll before uploading, rather than deciding alone, because a thumbnail flopped in February - informationChannel has 62,000 subscribers, the goal is 100,000 by the end of next year usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Creators%20%26%20influencers) Authors & novelists ## Your world never forgets again. Your AI knows every character, every relationship and every open plot thread. It reminds you exactly when it matters and ties the threads together more often and more elegantly. - Characters stay consistent: backstory, quirks, voice. - Open threads and foreshadowing get picked up again. - From the first prompt of every session, your AI knows your book, your style and the rules of your world, without re-reading the whole manuscript. - From plot brainstorming to editing: one memory for every stage. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **Claude**Built-in** › You: **Write the scene where Mara enters the lighthouse.** ◇ Built-in memory: 10 entries sent along › AI: Here's a draft, in third person and past tense, with plenty of interiority: Mara walked up from Dornsiel harbour to the lighthouse and stopped at the door. The harbour master had shown her the way. Below her lay the sea, grey and still. I used the Dornsiel lighthouse because it appears in your notes. Is that right, and where does the scene sit in the book? ✗ Sends Mara into the Dornsiel lighthouse. A different tower is meant. › You: **Write the scene where Mara enters the lighthouse.** Saved memories · always sent in full - Writes a coastal novel in third person, past tense. - The novel's main character is Mara. - Prefers scenes with little dialogue and a lot of interiority. - Is writing a short story "The Last Ferryman" for the Nordlicht anthology, due 15 Dec. - Harbour master Brandt from Dornsiel speaks Low German, minor character from chapter 4. - Chapter 4 is set in Dornsiel harbour, Mara watches the lighthouse from there. - Working title of the novel was "Nebelküste". - Beach scene in chapter 5 cut. - Writes 90 minutes in the morning, goal 800 words a day. - Uses Scrivener on the laptop. helps a littlesent for nothing With Hebby **Claude**With Hebby** › You: **Write the scene where Mara enters the lighthouse.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: Here's the scene for chapter 9, approx. 2,500 words, dusk, storm rising. The opening and the ending: The door on the sea side stuck. No one had opened it since 1987. Mara put her shoulder to it. "One hundred and twelve steps. Who builds something like this?" She counted quietly and, from step 60, stopped looking down. At the top, the wave sign on the iron chest's lock matched her brass key. In the envelope lay page 1 of her father's letter, torn off mid-sentence. You decided Mara finds only page 1 because Ilse keeps the second. That's why the scene ends mid-sentence in the letter. Shall I sketch the scene with Ilse at the kitchen table next? ✓ Mara's fear, the key from chapter 7 and your plan, all in one scene. › You: **Write the scene where Mara enters the lighthouse.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationMara Lenz has been afraid of heights since she fell from the ferry's mast aged nine - behaviorWhen afraid, Mara quietly counts steps and cracks dry jokes, she never cries in front of others - informationKap Wehr lighthouse: decommissioned since 1987, 112 steps, door on the sea side sticks - eventChapter 7: Mara finds a brass key with a scratched wave sign in her father's boat, the lock is still unknown - decisionMara finds only page 1 of her father's letter in the lighthouse, rather than the whole letter, because Aunt Ilse keeps page 2 until chapter 10 - informationChapter plan is on Drive: Novel/Kap_Wehr/Chapter_plan.docx Do I need the chapter plan? Yes, the tone and length of the scene are only in there. Chapter_plan.docx CHAPTER PLAN The Light of Kap Wehr As of 2 Nov Ch 7 The Brass Key done Ch 8 Row with Aunt Ilse done Ch 9 Mara in the lighthouse open Time: dusk, storm rising Length: approx. 2,500 words Tone: quiet, tension from the narrow space End: quiet, letter breaks off mid-sentence Ch 10 Mara confronts Ilse open - informationCharacter sheet and family tree are on Drive: Novel/Kap_Wehr/Characters.docx Do I need the character sheet? No, Mara and her father are clear to me. Characters.docx stays closed - informationChapter 10: Mara confronts Aunt Ilse at the kitchen table on the stormy night, Ilse evades and makes tea - informationEditor Ms Sander criticised too many adjectives in chapter 3, the manuscript is due on 31 January usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Authors%20%26%20novelists) Agencies & teams ## One team memory. The right context for every client. Hebby is the shared memory of your team and all your agents. Whoever works on a client with AI gets the essentials right away: agreements, rules, how the process works and where the documents are. And sees only what they should. - Whatever was once discussed with the AI, the whole team knows from then on. - Every client's brand rules apply automatically. - Permissions per team and agent: client knowledge stays neatly separated. - No knowledge is lost when someone goes on holiday or leaves. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **Claude**Built-in** › You: **I'm taking over Café Morgenrot from Lisa, who's on holiday. Write three Instagram posts for the autumn campaign.** ◇ Built-in memory: 9 entries sent along › AI: Sure, three drafts for Café Morgenrot, with tone and structure like Hoffmann: 1. "Autumn is here: try our new pumpkin menu, 10% off until the end of the month. 🍂" 2. "Freshly baked for you: our autumn pastries, every day from 8 am." 3. "Last week of the autumn campaign, now with 10% off!" Does the tone fit? Otherwise give me the details of the campaign. ✗ Takes the formal tone and 10% discount from Bäckerei Hoffmann. Neither fits Morgenrot. › You: **I'm taking over Café Morgenrot from Lisa, who's on holiday. Write three Instagram posts for the autumn campaign.** Saved memories · always sent in full - Works at a small social media agency in Leipzig and looks after hospitality clients. - Colleague Lisa Frey also looks after several of the agency's hospitality clients. - Writes posts for Bäckerei Hoffmann in a formal Sie tone, contact Frau Brandt. - Bäckerei Hoffmann has an autumn campaign in October with pumpkin bread, 10% off, posts on Mondays and Thursdays. - At Bar Nachtigall, Herr Vogel approves posts via WhatsApp. - Writes the monthly newsletter for the Vital fitness studio, contact Herr Dreher. - Uses Unsplash stock photos when the budget is tight. - Wrapped up the summer campaign for the ice cream parlour Eisdiele Roma. - Prefers answers as bullet lists instead of prose. helps a littlesent for nothing With Hebby **Claude**With Hebby** › You: **I'm taking over Café Morgenrot from Lisa, who's on holiday. Write three Instagram posts for the autumn campaign.** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: Here are the three posts for the autumn campaign (12 to 25 October), as Lisa agreed them with Mr Sommer on 5 October: 1. Mon 12 Oct, launch: "Autumn has a taste now: oat pumpkin chai, €3.80 instead of €4.20. Only until 25 Oct." Photo: a steaming cup by the window. 2. Tue 20 Oct: "Student Tuesday! Show your ID and grab a cinnamon roll for €2.20 instead of €2.60." Photo: fresh trays behind the counter. 3. Sat 24 Oct, 9 am: "Last chance for the pumpkin chai. It's over tomorrow." All casual, nothing with "coffee indulgence", five hashtags at most. Lisa and Mr Sommer decided to use only real café photos, because stock photos don't land with the audience. So every post comes with a photo idea for your team. Shall I prepare the approval email to Mr Sommer? He needs it 3 days ahead, so by Friday 9 October for the launch. ✓ Finished posts based on what a colleague on holiday agreed. No questions. › You: **I'm taking over Café Morgenrot from Lisa, who's on holiday. Write three Instagram posts for the autumn campaign.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationCafé Morgenrot in Leipzig, owner Mr Sommer, Instagram audience mostly students - decisionPosts in a casual, first-name tone and without the words "coffee indulgence" and "delicacy", because Mr Sommer finds them kitschy - eventLisa agreed the autumn campaign with Mr Sommer on Monday 5 October, minutes on the Drive: Clients/Morgenrot/Autumn_campaign_minutes.docx Do I need the minutes? Yes, period, prices and post dates are only in there. Autumn_campaign_minutes.docx MINUTES Autumn campaign Café Morgenrot Mon 5 Oct, Lisa Frey / Mr Sommer Period Mon 12 Oct to Sun 25 Oct Main item Oat pumpkin chai €3.80 instead of €4.20 Student Tuesday (Tue, with ID) Cinnamon roll €2.20 (was €2.60) Posts 3: start Mon 12 Oct, Tue 20 Oct, Sat 24 Oct (last chance) Photos real ones from the café only Approval Mr Sommer, by email, 3 days before publishing - decisionOnly real photos from the café, not stock photos, because the audience finds stock photos fake - behaviorCafé Morgenrot's posts do best on Saturdays at 9 am, more than 5 hashtags hurt reach - informationMr Sommer approves all posts by email, at least 3 days before publishing - informationCafé Morgenrot's brand guide is on the Drive: Clients/Morgenrot/Brand_guide.pdf Do I need the brand guide? No, I already have the tone and the list of banned words. Brand_guide.pdf stays closed - informationCafé Morgenrot opens a second location in Halle in spring usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Agencies%20%26%20teams) Students & researchers ## Your agent knows your research. Even what isn't published yet. Your papers, your findings and connections only you know because they're brand new: with Hebby, your agent keeps it all in mind. It gives smarter advice and links new findings to old ones the moment you talk about them. - From the first prompt of every session, your AI knows what your work is about and which papers matter. Without you explaining it, without it re-reading everything. - Status and decisions stay present, even when you switch between projects. - Fewer questions, more links between new and old findings. - In the lab, in the library, on your phone: always the same status. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **New result on the word-pair test: significant only in group B (t(30) = 2.90, p = .007), not in group A (p = .32).** ◇ Built-in memory: 9 entries sent along › AI: Exciting, for your psychology dissertation this sounds like a clear result. In R I would report both groups separately with t-tests, as in your pilot study: B significant, A not, which points to a real difference. Is this part of the follow-up study with Prof. Winkler? What exactly are groups A and B, and how large are the samples? ✗ Your field and tool are known, your study is not. The interpretation comes from the old pilot study. › You: **New result on the word-pair test: significant only in group B (t(30) = 2.90, p = .007), not in group A (p = .32).** Saved memories · always sent in full - Is writing a dissertation in psychology. - Mostly analyzes data in R. - Plans a follow-up study on stress and working memory with Prof. Winkler. - Analyzed the pilot study on stress and memory (n = 24) in 2023, both groups separately with t-tests. - Summarized Albrecht et al. (2022) on sleep deprivation and attention in spring. - Submits an abstract for the DGPs conference in December. - Sent the methods chapter to Prof. Winkler in September, feedback incorporated. - Runs a statistics tutorial for first-year students on Wednesdays. - Prefers answers in German. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **New result on the word-pair test: significant only in group B (t(30) = 2.90, p = .007), not in group A (p = .32).** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: That fits Hartmann & Voss (2024), but don't write it up as “effect only with sleep” yet. 1. Size: t = 2.90 with n = 31 gives d ≈ 0.52. Hartmann & Voss found d = 0.48 (CI 0.15 to 0.81). So your B is right in the expected range. That B has only 31 instead of 50 people because of the lab closure in May widens the interval but doesn't change the finding. 2. Group A already started at 74 % on the pre-test. Above about 80 % there is hardly any room to improve, so the missing effect in A may be a ceiling. 3. You decided to test the group difference directly, because “significant in B, not in A” doesn't show a difference. So run group × time and report only after that. Shall I turn this into the one-pager for Prof. Winkler? ✓ A proper assessment instead of questions. With your study, the right paper and your analysis plan. › You: **New result on the word-pair test: significant only in group B (t(30) = 2.90, p = .007), not in group A (p = .32).** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationStudy “Sleep and word-pair learning”: group A 20 minutes of rest (n = 52), group B 90-minute nap (n = 31), word-pair test before and after the break; pre-test means A 74 %, B 58 % - decisionTest the group difference directly (group × time), not each group separately, because “significant in B, not in A” alone doesn't prove a difference between the groups - eventSleep lab closed in May for refurbishment: group B has only 31 instead of the planned 50 people - behaviorOn the word-pair test, a pre-test score above about 80 % leaves hardly any room for improvement - informationHartmann & Voss (2024) on napping and word-pair learning is in Zotero with notes: collection Sleep/Consolidation Do I need the paper? Yes, I need their effect size for comparison. Hartmann_Voss_2024.pdf Hartmann & Voss (2024) Napping and word-pair learning, Exp. 1 3.2 Results (N = 40, within-subject) Nap, 90 min: gain +11.4 pp, d = 0.48 95 % CI [0.15, 0.81] Rest, 20 min: gain +2.1 pp, d = 0.11 Condition × time: F(1, 39) = 5.62, p = .023 p. 9: pre-test above 80 % were excluded (ceiling effect). - informationProf. Winkler wants results before every meeting as a one-pager with a figure, effect size and confidence interval - informationRaw data and analysis script are on the institute drive: Studies/Sleep/Analysis_v3.R Do I need the script? No, you've already given me the key figures. stays closed - decisionFocus the introduction on sleep spindles, not dream sleep, because only for those is there enough word-pair literature usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Students%20%26%20researchers) Teachers & lecturers ## Your agent knows your teaching. Curriculum, courses, material, what worked and what didn't: Hebby remembers. Your agent plans lessons that pick up where you left off, with material that fits. - Lesson plans that build on the last lesson. - Material matched to each class's level and pace. - Ideas that once took off come back to the table. - Back after the holidays or in a new school year? Your AI briefly explains your own teaching concept and why you built it that way. It knows where each piece of material is, what worked and who contributed what. - Hebby remembers content and methods, not data about your students. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **Plan the next biology lesson for 8b.** ◇ Built-in memory: 9 entries sent along › AI: Sure! As a biology teacher at a secondary school you get the plan as a table with timings. I'll take photosynthesis as the topic, as with 8a: 15 minutes teacher presentation, 20 minutes worksheet, 10 minutes consolidation. Does that fit 8b, or are they further along? Tell me briefly what was covered last, then I will adjust it. ✗ Subject and format are known, 8b is not. The suggestion copies the lesson for 8a. › You: **Plan the next biology lesson for 8b.** Saved memories · always sent in full - Teaches biology and chemistry at a secondary school. - Likes lesson plans as a table with timings. - Biology 9a: genetics unit, Mendel's laws finished in October. - Biology 8a: taught photosynthesis in spring as a teacher presentation with a worksheet. - Chemistry 10c: class test on acids and bases on 13 November. - Planned a field trip to the city forest for 7a in May, done. - Parents' evening for 6c on 17 November, prepare the class trip contribution. - Signed up for the training course “Digital media in subject teaching” in Hildesheim in January. - Prefers answers in German. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **Plan the next biology lesson for 8b.** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: For Monday, 9 November (photosynthesis 1 in the unit plan), I suggest: 1. A 12-minute group rally on respiration. The quiz was weak, and 8b concentrates better after group work. Focus on four station cards: don't mix up respiration and fermentation. 2. 25 minutes introducing photosynthesis, still without a slide lecture. You decided to introduce processes with an experiment, because 8b won't retain them otherwise. The pondweed experiment is on Thu 12 November in lab B14, so on Monday we prepare: write down a hypothesis, a sketch, the reaction equation as a gap text. 3. 8 minutes to consolidate: what does the plant need, what is produced? Shall I prepare the practice sheet for the class test on 26 November as well? ✓ A lesson that picks up from the last one. With your plan, your room and your methods. › You: **Plan the next biology lesson for 8b.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationBiology 8b: 2 lessons a week (Mon, Thu), unit “Cell metabolism”; cellular respiration finished, in the final quiz the class often mixed up respiration and fermentation - behavior8b concentrates noticeably better after group work - decisionIntroduce photosynthesis with an experiment, not slides, because 8b only retains abstract processes when they can see them - eventMs Kaya from the biology collection is providing pondweed and lamps in lab B14 for Thu 12 November - informationUnit 3 plan for 8b is in the teachers' folder: Biology/8b/Unit3_Plan.docx Do I need the unit plan? Yes, it has the lesson sequence and the test. Unit3_Plan.docx YEAR PLAN BIOLOGY 8b, Unit 3 Cell metabolism (Mon + Thu) Mon 2 Nov Respiration 2, quiz Mon 9 Nov Photosynthesis 1: overview Thu 12 Nov Photosynthesis 2: experiment (lab B14) Mon 16 Nov Photosynthesis 3: leaf structure Thu 19 Nov Revision, practice sheet Mon 23 Nov Buffer Thu 26 Nov Class test (45 min) Mon 30 Nov Unit 4: ecosystems - informationFor every class test, 8b gets a practice sheet with solutions a week in advance - informationQuiz questions and solutions on respiration are in Biology/8b/Quiz_Respiration.pdf Do I need the quiz? No, the weak spot is already clear. Quiz_Respiration.pdf stays closed - event8b goes to the university student lab in spring usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Teachers%20%26%20lecturers) Job seekers ## Every application in view. Every interview in mind. Hebby remembers where you applied, with which cover letter, who got back to you when and what mattered in each interview. Your AI helps with purpose instead of starting from zero every time. - All applications, deadlines and contacts in one place, no spreadsheet needed. - Cover letters that fit your profile and the company, without re-uploading your CV. - Your AI reminds you when to follow up. - Interview prep with everything you already know about the company. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **I have the interview at Solbrink tomorrow.** ◇ Built-in memory: 9 entries sent along › AI: Good luck for tomorrow! As a project manager in the energy sector you surely have good examples. I would prepare you for questions on team leadership, as at Helion Systems, and for salary I'd name a range like at Nordwerk. Feel free to send me the ad. Which position is the interview for exactly, and how will it run? ✗ Industry and job are known, your application is not. The advice comes from older applications. › You: **I have the interview at Solbrink tomorrow.** Saved memories · always sent in full - Works as a project manager in renewable energy. - Is looking for a new job in project management. - Applied to Nordwerk Energie in July, rejected in August. - Had an interview at Helion Systems (team lead wind farms) in October, feedback still pending. - Applied to Enerplan GmbH as photovoltaic project lead in September, rejection by email. - Revised the CV in English in September. - Earned the PMP certificate in March 2024. - Named the salary expectation at Nordwerk Energie as a range of €58,000 to €66,000. - Prefers answers in German. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **I have the interview at Solbrink tomorrow.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: Then let's prepare you now: tomorrow, 6 November, 10 am, 60 minutes with Dr Seidel and Mr Albers. According to the ad, Solbrink wants Spanish, grid connection experience and budget responsibility up to €3m. You have Zaragoza for that: 3.2 MW, €2.1m, three weeks of delay at the grid connection, caught up again. That's your answer to “a difficult project” and to the weakness question. “I take on too much” sounded like a cliché in the practice run. On salary: you decided to name a fixed figure, €62,000, because a range gives away your lower limit. And ask how the 30 % travel is shared across the team. Shall I draft the thank-you email for after the interview? ✓ Preparation for exactly this job. With the ad, your project and your salary decision. › You: **I have the interview at Solbrink tomorrow.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationSolbrink GmbH, position Project Manager Solar Parks Abroad, applied on 12 October; interview Fri 6 November, 10 am by video with Dr Seidel (HR) and Mr Albers (head of international projects) - eventPhone call with Dr Seidel on 22 October: the interview lasts 60 minutes, focus on client projects in Spain and Italy - informationLast employer Terra Montage: led the Zaragoza solar park 2023 to 2024, 3.2 MW, budget €2.1m, 3 weeks of delay at the grid connection, caught up again - behaviorWhen asked about weaknesses, “I take on too much” came across as a cliché in the practice run; the grid connection example in Zaragoza feels more genuine - decisionState the salary expectation as a fixed figure (€62,000), not a range, because a range gives away the lower limit - informationSolbrink's job ad is in the Drive: Applications/Solbrink/Job_ad.pdf Do I need the job ad? Yes, I need the exact requirements. Job_ad.pdf SOLBRINK GmbH, Hamburg Project Manager Solar Parks Abroad Permanent, full-time Your tasks - Lead solar parks in Spain and Italy (2 to 5 MW) - Budget responsibility up to €3m - Manage the contractors on site - Travel: approx. 30 % Your profile - Fluent negotiating Spanish - Experience with grid connection procedures - informationAfter interviews, send a short thank-you email the same day, follow up after 7 days without a reply - informationCover letter and CV for Solbrink are in the Drive: Applications/Solbrink/Cover_letter.docx Do I need the cover letter? No, I know its key points. Cover_letter.docx stays closed - eventA second application is running at Windfeld AG, reply expected there by 20 November usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Job%20seekers) Tinkerers & vibe coders ## All your agents. One memory. Full throttle. Hebby knows all your little projects, agents, helpers and tools. They all share one memory: each knows what you've built, what you're planning and what's next. - Every new session starts with knowledge instead of re-reading: your agent knows your projects from the first prompt. - Far fewer questions, far better suggestions, far fewer bugs. - The agent on your laptop knows what the one on your PC built yesterday. - Hebby captures your style and your ideas. Your AI increasingly runs with them on its own. - Smart suggestions that connect your projects. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **Codex**Built-in** › You: **Finish my Discord bot, I want to show it to my friends tonight.** ◇ Built-in memory: 9 entries sent along › AI: Sure, I'll help with your Discord bot. I assume you mean Pixel for the server "Spielabend", where /umfrage is still missing. I'll finish that tonight, in Python with discord.py. I can only guess which bot it is, though, and I don't know whether other commands are missing. Is that right, or tell me the name and repo. 😄 ✗ Guesses the wrong bot (Pixel instead of Moppel) and builds /umfrage instead of /wetter. › You: **Finish my Discord bot, I want to show it to my friends tonight.** AGENTS.md · always sent in full - - Builds hobby projects in Python, repos under ~/projekte - - Has Discord bots for friends, using slash commands - - Second bot "Pixel" for the server "Spielabend", repo ~/projekte/pixel, command /umfrage still open - - Old bot "Würfelmeister" built with discord.js (Node 16), no longer maintained - - Smart home dashboard with Flask, sensors via MQTT, start with python app.py - - Uses VS Code on the laptop, git commits in German - - Never put secrets in the repo, .env is in .gitignore - - Likes short answers with emojis - - Wanted to try a music feature for Pixel later, using yt-dlp helps a littlesent for nothing With Hebby **Codex**With Hebby** › You: **Finish my Discord bot, I want to show it to my friends tonight.** 9/11 memories fitting · 6 useful, of them 6 used · 1 anticipated › AI: Moppel is done except for /wetter, which I'm building now in ~/projekte/moppel. 1. I'm reusing the lookup from your smart home project: wetter(ort) from open-meteo, no key, 10-minute cache. Dortmund is the default, and you can pass a place as an option. On a timeout, Moppel says in a friendly way that it isn't working right now. 2. The reply is casual with temperature, rain and wind, like "14 °C, dry, 12 km/h wind 🌤". 3. The token stays in .env, as since the September incident. I'm not using match-case, because the Pi only runs Python 3.9. 4. So /wetter is there tonight, I only sync for "Kellerkinder". Globally it takes up to an hour. You decided Moppel should run on the Raspberry Pi, because your laptop is often off in the evening. So I'm offering to deploy him there once the test passes. Shall I? ✓ Your bot gets finished, with the weather lookup from another project. No questions. › You: **Finish my Discord bot, I want to show it to my friends tonight.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationDiscord bot "Moppel": Python with discord.py, repo ~/projekte/moppel on the laptop, /würfel and /witz done, only /wetter is missing - eventWeather lookup built in the smart home project, notes are in ~/projekte/smarthome/NOTIZEN.md Do I need the smart home notes? Yes, they say how the weather lookup is called. NOTIZEN.md NOTES Smart home wetter.py - fetches data from open-meteo - no API key needed - wetter(ort) returns a dict: temp, regen, wind - default: Dortmund 51.51 / 7.47 - cache: 10 minutes - on timeout: returns None Open: place names instead of coordinates - behaviorNew slash commands only show up in Discord after tree.sync(); globally that takes up to an hour, for a single server it's instant - informationMoppel is tested on the server "Kellerkinder" (7 friends), tone casual, an emoji is welcome - eventBot token was accidentally on GitHub on 3 September, regenerated, since then it only lives in .env - behaviorThe Raspberry Pi only runs Python 3.9, match-case doesn't work there - decisionRun Moppel permanently on the Raspberry Pi instead of the laptop, because the laptop is often off in the evening - informationIdea list for Moppel is in ~/projekte/moppel/IDEEN.md Do I need the idea list? No, today is only about /wetter. IDEEN.md stays closed - informationIdea for later: Moppel announces the bin collection dates usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Tinkerers%20%26%20vibe%20coders) Dungeon masters & tabletop RPG ## Every character, every promise, every unpaid debt. Your campaign has been running for months? Hebby remembers characters, places, open quests and everything your party got up to. Your agent improvises with you without breaking canon. - From the first prompt of every session, your AI knows the campaign, without re-reading all the session logs. - NPCs stay true to themselves across dozens of sessions. - Open threads return at just the right moment. - Session prep in minutes instead of hours. - Your players marvel at how much your world remembers. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **Claude**Built-in** › You: **The party enters the tavern in Ravenford.** ◇ Built-in memory: 10 entries sent along › AI: You enter the tavern in Ravenford. Smoke hangs in the air, a bard tunes his lute, and in the corner sits a stranger who wants to offer you a quest. Innkeeper Egon Brandt steps out from behind the bar and demands the 40 gold for the furniture, while the watch nods at you kindly. I had Hohenfels in mind, does that work for Ravenford too? ✗ Turns Egon from Hohenfels into Ravenford's innkeeper. The debt runs the other way. › You: **The party enters the tavern in Ravenford.** Saved memories · always sent in full - Runs a D&D 5e game with four players, sessions every two weeks. - The party consists of a fighter, a rogue, a cleric and a wizard. - Likes to include a stranger offering a quest in taverns. - Tavern "The Golden Griffin" in Khazar: landlady Samira, night market outside. - Innkeeper Egon Brandt from Hohenfels: the party owes him 40 gold for broken furniture. - The Hohenfels city watch has been friendly to the party since session 8. - The dragon of Grauwacht was defeated in session 5, closed. - Campaign villain is the necromancer Valdrik, appearance planned for session 15. - Critical hits double the damage dice. - Prefers atmospheric descriptions, preferably short. helps a littlesent for nothing With Hebby **Claude**With Hebby** › You: **The party enters the tavern in Ravenford.** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: The door of "The Crooked Raven" swings shut, and the laughter in the taproom dies. Bodo nearly drops his mug and immediately starts talking about the weather. The 50 gold he's owed you since the giant rats, he'd rather not mention. For you as GM: 1. A wanted poster hangs by the bar: 20 gold on Lyra's head. The doorman recognises her by her scar as soon as her hood falls. 2. Edda's amulet glows softly. A cultist sits in the corner. 3. Brokk's axe has been ready at Gunnar's since Monday. You decided Lyra's theft has consequences because her player wanted that. So the watch won't look away. Shall I prepare the cellar with the smugglers' tunnel as an escape route? ✓ Your innkeeper, your debts and the fallout of night 12, no re-reading. › You: **The party enters the tavern in Ravenford.** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - informationRavenford: harbour town on the Mist River, tavern "The Crooked Raven", innkeeper Bodo Kessel, city watch under Captain Varga - eventBodo owes the party 50 gold since they cleared his cellar of giant rats in session 6 - behaviorBodo dodges and talks about the weather as soon as money comes up - decisionLyra's theft has consequences: the watch keeps hunting her, rather than forgetting it, because her player wanted consequences for her actions - eventSession report for night 12 is on Drive: Campaign/Reports/Night_12.txt Do I need the report? Yes, I only know the state of the watch, the axe and the amulet from there. Night_12.txt SESSION REPORT Night 12, 30 Oct Place: Ravenford, harbour district, night - Lyra steals Orsino's signet ring, the watch chases the party - Escape over the rooftops, the watch loses them at the fish market - Brokk's battleaxe is with smith Gunnar, ready in 3 days - Edda's amulet glows near cultists - Captain Varga puts 20 gold on Lyra's head Open: signet ring still with Lyra - informationCity map of Ravenford is on Drive: Campaign/Maps/Ravenford.png Do I need the city map? No, the tavern is the setting, routes don't matter tonight. Ravenford.png stays closed - informationAn old smugglers' tunnel to the harbour ends in the cellar of "The Crooked Raven", Bodo knows nothing about it - decisionTavern brawls end after 3 rounds because the watch steps in then, rather than playing them out forever usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Dungeon%20masters%20%26%20tabletop%20RPG) Self-organisation & ADHD ## Your AI knows again what you set out to do. Promises, deadlines, half-baked ideas and unfinished plans you talked through with your AI: Hebby remembers them all. Ask what's still open and your AI knows. If your calendar is connected, it adds deadlines right away. - Last week's promises are not forgotten. - Unfinished things come back to the table before they fizzle out. - Jump around the way your mind does. Your AI jumps along. - Told on your phone, continued on your laptop: same status. ### ▶**See the example**Built-in memory and Hebby side by side With built-in memory **ChatGPT**Built-in** › You: **What was I planning to do this week?** ◇ Built-in memory: 9 entries sent along › AI: I'm not sure. From memory I have: check the utility bill from Brandt, tax documents for Mr Weber and the annual plan for Mr Okafor by 20 November. Whether that is still due this week, I cannot say. Write down what is on your plate right now, and I will turn it into a weekly list with the important tasks in the morning. ✗ Your preferences are known, your deadlines are not. It names tasks that were long since done. › You: **What was I planning to do this week?** Saved memories · always sent in full - Likes to plan with weekly lists. - Prefers to focus in the morning. - Wanted to check the utility bill from the Brandt property management by the end of October. - Promised Mr Okafor the annual plan for the support association by 20 November. - Sent 2024 tax documents to Mr Weber (old tax office) in May, done. - Signed up for the Excel course at the adult education centre by 15 October, done. - Submitted the expense report for the Cologne business trip in October. - Wanted to start the Spanish course at the adult education centre in September, not started yet. - Prefers short answers. helps a littlesent for nothing With Hebby **ChatGPT**With Hebby** › You: **What was I planning to do this week?** 8/10 memories fitting · 5 useful, of them 5 used · 1 anticipated › AI: Two things are heading for a deadline this week: 1. Today, 6 pm: sign up for the photography course “Light and Portrait” (€139, only 12 places). That takes ten minutes. You decided to do the nearest deadline first for things you've started, so the course comes before the tax documents. 2. Tomorrow: the tax documents for Ms Lenz, which you promised her. Your payslip summary and donation receipts have been with her since Tuesday. According to her checklist, four are still missing: the dental invoice (€840), the tradesman invoice for the bathroom, the pension plan statement and your number of home office days. Paperwork doesn't work for you after 8 pm, so gather the receipts early today. Shall I prepare the scan order and a short cover email to Ms Lenz? ✓ Your week in two paragraphs. With deadlines, promises and what's still missing. › You: **What was I planning to do this week?** Inside Hebby · behind the scenes Your AI sees none of this. It only gets the result. - eventPromised Ms Lenz (tax adviser) the tax documents by Fri 6 November; payslip summary and donation receipts already sent on 3 November - informationSign-up for the photography course “Light and Portrait” at the adult education centre: €139, 12 places, registration closes Thu 5 November, 6 pm - decisionFor things started, do the nearest deadline first, not by importance, because limited places and deadlines would otherwise expire - behaviorPaperwork doesn't work after 8 pm, so do it earlier - informationMs Lenz's checklist arrived on 28 October as an email attachment: inbox, email “Documents tax 2025”, attachment Checklist_Tax_2025.pdf Do I need the checklist? Yes, I need to know what is still missing. Email attachment: Checklist_Tax_2025.pdf DOCUMENTS INCOME TAX 2025 Lenz Tax Office [ ] Annual payslip summary 2025 [ ] Donation receipts (summary) [ ] Dental invoice of 14 March (€840) [ ] Tradesman invoice, bathroom (Oct) [ ] December statement (pension plan) [ ] Number of home office days Deadline: Friday, 6 November - informationMs Lenz prefers receipts as a single PDF in the order of her checklist - informationProgramme for the photography course came as an email attachment: Photo_Course_Programme.pdf Do I need the programme? No, this is only about the registration deadline. Photo_Course_Programme.pdf stays closed - informationHalf-formed idea: convert the balcony into a raised bed in spring, materials list still open usedanticipatedfitting **Anything missing for your field?** Your workflow looks different or you have questions? Write to us, we are happy to add to it. [Write to us →](https://www.hebby-memories.com/en/contact/?bereich=Self-organisation%20%26%20ADHD) ## Your role isn't listed? No problem. Hebby learns from any kind of work, whatever pays your bills or fills your weekends. Give it a try. Early Access ## Don’t worry, be Hebby. An agent that remembers. Hebby is now rolling out to our first users. Tell us briefly what you work with. [Request Early Access](https://www.hebby-memories.com/en/early-access/)