Opportunities › Sandboxed agent infrastructure for non-engineering teams › Proposed app
Penfold
Safe sandboxes so PMs, support and finance can use coding agents on real repos
- Score
- 43
- confidence 64
- Evidence
- 15
- items on the page
- Rivals
- 8
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop and mobile
- PWA: on computers and phones alike
- Market since
- 2021
- Emerging market · AI estimate
ScreensConcept
Desktop and mobile: Engineers set up repos and review diffs on desktop while support and finance teams start tasks from Slack on their phones, requiring both platforms for full workflow. Sample screens on a computer and a phone, in the proposed brand. Their numbers are sample data.
New task
Sandbox starting- Requested by
About this appAI proposal
Penfold lets engineering teams hand coding agents to the rest of the company without handing over the keys. An engineer sets up a repo once: instructions, skills, scoped integrations and the full environment. Penfold snapshots it and runs every task in a fresh, isolated sandbox.
Product managers, support, finance and on-call staff start tasks from Slack, Linear or GitHub. Nothing merges until a person reviews the diff and a live preview. Secrets stay behind a proxy the agent never sees, and command and network rules are enforced outside the sandbox.
It differs from a home-built setup because the environment, guardrails and shared know-how are maintained for you. It works with the agent you already use and the sandbox provider you choose. Pricing is a subscription plus compute, with no markup on model calls.
- Tagline
- Let everyone use coding agents. Keep the codebase yours.
- Voice
- CalmPrecisePractical
- Why this name
- A penfold is a small enclosure that keeps things safely contained. That is what each sandbox does for an agent.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Environment snapshots
Capture Docker Compose, databases and services once. Every task starts from the same working setup, with no local install for anyone.
No member yet
Start from Slack, Linear or GitHub
Non-engineers file a request where they already work. The agent picks it up and replies in the same thread with results.
No member yet
Shared skills and instructions
Engineers write the context once: prompts, repo conventions and skills. Every agent run gets the same knowledge.
No member yet
Secrets behind a proxy
Agents call services through a managed proxy that injects credentials. The sandbox never holds a raw key.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#12403CDeep Teal · primary#F6F3ECPaper · surface#1B1F1EInk · text#E8A33DSignal Amber · accent#5E8C6AMoss · data#B54A3ABrick · warning
Typography
Aa
Penfold
PricingAI proposal
The material gives no price for this kind of product. E2B's pricing page lists a free plan and one paid plan at $150 a month; Cursor lists $20 to $40 a month. Tiers are a starting point to test, plus compute at cost.
Free
Free
- 1 repo
- 3 users
- Pay for compute used
- Community support
Team
$150 a month
- 10 repos
- Unlimited requesters
- Slack, Linear, GitHub triggers
- Command and network rules
- Secrets proxy
Enterprise
Contact us
- SSO and audit export
- Bring your own sandbox account
- Custom limits
- 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
- Partners and integrations. Slack, Linear and GitHub apps put Penfold in front of every teammate who requests work. Requesters see the bot, and engineers see the review links in pull requests.
- Do they become customers?
- Requesters are often non-buyers; the buyer is the engineering lead. Test whether bot users in a team prompt that lead to start a paid plan, or only use it.
- First customers
- Engineering leads at startups with 20 to 200 staff who already pay for a coding agent and want support or finance staff to use it.
- Where to reach them
- Hacker News: Show a demo of a support fix going from Slack request to reviewed pull request, and ask for design partners.
- Founder and CTO communities: Offer a free setup in startup Slack groups to teams that built their own gating.
- Direct outreach: Message engineering leads who have posted about agent rollouts or unmergeable pull requests.
- First milestone
- Five engineering teams pay for a team plan after a two-week trial with at least three non-engineers each.
Why people already pay for this
- VerifiedHacker News post received 103 points and 30 comments
- Public claimWhen trying to use coding agents across a whole team, most PRs are unmergeable, repos require one-off local setup, and skills live in one person's head.
- Public claimNon-engineers cannot safely touch a real codebase when using agents without risking a bad deploy or a secrets leak.
- Public claimOne customer built an on-call inspector agent wired to PagerDuty, Sentry, and their repo that finds alert causes and opens PRs with unit tests before anyone gets paged.
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 · Sandbox runtimeSandbox providers such as E2B or DaytonaIsolated environments with snapshots are the core of the product and should be rented, not built.
- This app · Secrets proxyEgress proxy with credential injectionKeeps raw keys out of sandboxes and enforces network allow-lists outside the agent.
RisksAI inference
- PlatformAgent vendors or sandbox providers could add team controls of their own and make a separate layer less needed.
- TechnicalSnapshotting varied setups (Compose, databases, external services) reliably is hard, and each odd repo can cost support time.
- CommercialA similar product already claims named customers in the material. Differentiation must be proven with a narrow group.
- AIEven with review, agents can produce poor changes. If most pull requests stay unmergeable, non-engineers will lose trust.
- DataCustomers' source code and business data pass through the sandboxes, so security reviews will be demanding and slow sales.
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.