Opportunities › AI code completion and review for developers › Proposed app
Pocketbench
Local-first AI coding for VS Code, with fallbacks when a model is slow or limited
- Score
- 55
- confidence 66
- Evidence
- 24
- items on the page
- Rivals
- 6
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2022
- Emerging market · Verified
ScreensConcept
Desktop app: Developers spend long sessions in VS Code writing and reviewing code, requiring dense configuration screens and keyboard-driven workflows. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Fallback chain
All routes healthyAbout this appAI proposal
Pocketbench adds code completion, chat, edits and review to VS Code, and runs them on models on your own machine first. If you want more power, you can add a hosted model of your choice. Your code stays local unless you say otherwise.
The companion app (a PWA) is where you set the rules: which model answers which task, what may leave your machine, and a monthly spend cap. When one model is slow or limited, requests move to the next one on your list.
It is for developers who want AI help without a locked-in subscription, and for small teams that want to see what their code assistant costs and where their code goes.
- Tagline
- Your code, your models, your limits.
- Voice
- PlainCalmPractical
- Why this name
- A bench is where a developer works and tests tools. Pocket suggests it fits on your own machine and travels with you.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Local models first
Connect a model runtime on your machine or network. Completion, chat and edits run there by default, with no account needed.
No member yet
Fallback chain
Order your models: local, your own hosted key, then managed. When one is slow or limited, the next answers, and you see which one did.
No member yet
Privacy rules per folder
Mark folders or file patterns as local-only. Those files never reach a hosted model, and the request log shows it.
No member yet
Review before you push
Run a review pass on a diff or branch. Comments appear inline in VS Code and as a checklist in the app.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#14342BDeep Pine · primary#F6F1E7Paper · surface#1B2A24Ink · text#E8A33DAmber · accent#2F6B4FMoss · data#B5472EClay · warning
Typography
Aa
Pocketbench
PricingAI proposal
Based on GitHub Copilot's pricing page (VERIFIED): a free plan and two paid plans from $4 to $21 a month. These tiers sit inside that range and are a starting point to test.
Local
Free
- Local models in VS Code
- Completion, chat, edits
- Privacy rules
- Bring your own keys
Pro
$8 a month
- Managed hosted fallback
- Spend cap and usage log
- Model comparison bench
- Branch review reports
Team
$20 a month
- Per seat, shared rules
- Team usage dashboard
- Shared privacy policies
- Priority support
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Branch review reports can be shared as a link in pull requests. Teammates who open them see the findings and which model produced them, with a link to Pocketbench.
- Do they become customers?
- Reviewers are developers and could install it, but a link opened once is exposure only. Test how many report viewers install the extension and how many then upgrade.
- First customers
- Freelance and small-team developers using VS Code who pay for an AI coding subscription and care about privacy or rate limits, in teams of 1 to 10.
- Where to reach them
- Hacker News: Post a Show HN focused on local-first completion with automatic fallback, and answer questions on rate-limit complaints.
- GitHub and VS Code Marketplace: Publish the free extension with a clear README so developers find it when searching for local AI assistants.
- Developer subreddits and Discords: Share benchmarks from the comparison bench in local-model communities.
- First milestone
- Five hundred installs of the free extension and twenty-five paying Pro users within 60 days of launch.
Why people already pay for this
- VerifiedGitHub repository gained 3665 stars in 1146 days
- Public claimThere is demand for AI-powered code completion and editing integrated into developer workflows.
- Public claimDevelopers seek alternatives to cloud-dependent AI coding tools, prioritizing privacy and infrastructure control.
- Public claimThe tool supports multiple deployment models: local inference and hosted options.
- AI inference
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · VS Code extensionTypeScript extension using the VS Code APICompletion, chat and inline review must run inside the editor; the PWA alone cannot do that.
- This app · Local bridgeSmall localhost service to local model runtimesLets the app and extension talk to models on the developer's machine without sending code out.
RisksAI inference
- TechnicalCompletion quality and speed on typical laptops with local models may feel worse than cloud tools. Weak first impressions would hurt adoption.
- CommercialA free local tier may convert few users to paid. Pro must offer a clear reason, such as fallback and spend control.
- PlatformThe product depends on the VS Code extension API and marketplace rules, which can change or limit features.
- CommercialOpen-source assistants already exist, and the material says nothing on how many developers would pay for routing and review.
- DataManaged hosting means handling user code on servers. Privacy promises must be matched by retention and logging limits.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.