Runtm
Runtime runs AI agents for payment operations in isolated cloud sandboxes that ask before they act.
If your agents touch money movement, Runtime is one of the few harnesses that treats approval, evidence, and per-agent scoping as the product rather than a bolt-on — the $12,400 return-of-funds flow with a drafted reply and evidence file is the clearest demo of what you're buying. The catch is scope: the homepage is payment ops, but the pricing page still sells coding agents, so confirm which story you're onboarding. Pick it when a partner bank or auditor will ask what the agent did; pass if
Verified 2h ago · liveness 73/100 · cite: rightaichoice.com/tools/runtm
- Payment operations teams tracing exceptions, partner-bank RFIs, and returns of funds before they become a backlog
- Risk, underwriting, and KYB teams that need an evidence packet attached to every reserve or hold decision
- Payments support teams answering "where's my money" questions with real ledger and processor data
- Compliance and security leads who need per-agent scoping, PII masking, and a searchable audit trail of every run
- Solo developers who want a local coding agent and don't need approvals, evidence packets, or role-based access
- Teams expecting a no-code managed automation product — you describe the agent, but you still own the SOPs and tool access
- Organizations needing fully offline, air-gapped deployment (the sources describe self-hosting and in-VPC, not air-gapped)
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Runtime if you're a solo developer who just needs a local coding agent — you'd be paying $99/seat for approval gates, scoped credentials and an audit trail you have no compliance team to hand them to.
The free tier's 500 credits are consumed by agent runs, so a single long investigation can burn through your evaluation budget before you've seen real output.
Runtime's pricing suits funded engineering and fintech ops teams, not individuals. The $0 Free tier covers evaluation only (500 credits, 1 session, 4 vCPU / 4 GB). Teams at $99/seat/month undercuts assembling your own orchestration plus sandboxing plus audit logging, and is cheaper than most enterprise agent-governance platforms that quote custom-only. Enterprise is custom-priced and effectively bundles self-hosting, SSO/SCIM and an SLA — comparable governance products from established security
In short
Runtm — Runtime runs AI agents for payment operations in isolated cloud sandboxes that ask before they act. Best for Payment operations teams tracing exceptions, partner-bank RFIs, and returns of funds before they become a backlog, Risk, underwriting, and KYB teams that need an evidence packet attached to every reserve or hold decision, Payments support teams answering "where's my money" questions with real ledger and processor data. Free to start; paid plans from $99/user/mo.
What people actually say about Runtm — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
30 mentions across 3 sources (Hacker News, Bluesky, GitHub) · researched Jul 6, 2026.
Average across the 3 sources that answered — each source counts once, not each post.
- +Agent-agnostic: works with Claude Code, Cursor, Codex, Gemini CLI.
- +Bring your own API keys (BYOK) reduces vendor lock-in.
- +Sandboxed environments with pre-warmed snapshots boot quickly.
- +Mission Control dashboard provides live visibility and spend controls.
- +Permission gates and allowlists enable governance for non-engineers.
- −Signup bugs (OAuth failure) can block new account creation.
- −Broken Discord invite hinders community support access.
- −Very limited real-world usage data—reliability unproven.
- −No large-scale performance benchmarks available publicly.
- −Documentation may be incomplete given early stage.
- • AI provider API costs (BYOK) – not included in Runtime pricing.
- • Enterprise self-hosting may require internal infrastructure investment.
Viability Score
How well maintained and how widely used is Runtm? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Run AI ops and coding agents on their own isolated computer in your cloud
- Approval gates for money movement, reserves, and replies to partner banks
- Per-agent role-based access — read-only ledger, approval-required returns of funds
- Call agents from Slack, Teams, email, SMS, or voice
- Answer a merchant support line with a voice agent
- Trigger runs from Splunk alerts or a daily schedule
- Connect APIs, databases, and MCP servers, plus a browser for portals with no API
- Every run stored with steps, tool calls, duration, cost, and evidence packet
- Agents learn from resolved cases and propose updates to their own skills
- One governed memory shared across support, payment ops, finance, and compliance
- PII, cardholder data, and secrets masked before any prompt or log
- Run open-weight models locally inside your network for sensitive work
- Self-host, run agent computers in your own AWS/GCP/Azure account, or bring your own sandbox
- Agent-agnostic: Claude Code, Codex, OpenCode, Gemini, GitHub Copilot and more
- SDK, CLI and API access for programmatic agent orchestration
About Runtm
Runtime is an agent harness for payment teams. Anyone on the team — payment ops, risk, compliance, underwriting, support — describes a job in plain English and attaches the SOP; Runtime builds the agent, gives it its own computer in your cloud, and runs it with approvals before anything moves money. The homepage demo is a banking ops agent that handles partner-bank RFIs, hold-harmless requests, and returns of funds, verifying against a ledger and drafting replies for human approval. The harness has four moving parts. Access control is per agent: bank email reads requests, the ledger is read-only, and returns of funds wait for a person. Triggers span Slack, Teams, email, SMS, voice, alerting systems, and schedules — a merchant support line can be answered by an agent. Tools come from your prompt and connect over APIs, databases, MCP servers, and a browser for processor and bank portals that have no API. Approvals cover money movement, reserves, and bank replies, and each run is stored with its steps, queries, cost, and evidence packet. Runtime treats every run as organizational memory. Agents learn from resolved cases and propose updates to their own skills, so a stuck payout investigated by support is already known to payment ops. Card numbers, SSNs, and secrets are masked before any prompt or log, and you can run open-weight models locally inside your network for the most sensitive work. Deployment is SaaS by default, with self-hosting, running agent computers in your own AWS, GCP, or Azure account, or bringing your own sandbox provider on Enterprise. The company is SOC 2 audited and states it never trains on your data. Pricing is published and starts free. The $0 tier includes 500 credits, one session, Claude Code, Codex and OpenCode, and a 4 vCPU / 4 GB sandbox. Teams runs from $99 per seat per month with unlimited sessions and users, all agents, shared templates and secrets, spend controls, RBAC, and priority support. Enterprise adds SAML SSO, SCIM, audit
Behind the Verdict
Runtime's pitch narrowed from general agent infrastructure to payment operations, and that focus is the reason to consider it. A banking ops agent that reads a partner-bank RFI, checks the ledger, recommends a reserve, and waits for approval is a specific, defensible job — not a demo of a chatbot. If your team spends its week on payout exceptions, chargebacks, KYB re-reviews, and "where's my money" escalations, this is built for exactly that queue. What we like is the shape of the controls. Permissions attach to the agent, not to a shared service account: read on the ledger, approval required for returns of funds. Approvals carry the evidence packet with them, so the person clicking Approve sees the queries and the reasoning. Runs are stored with cost and outcome — the demo shows $0.31 for a 4m 02s run with 14 tool calls. That is the artifact your compliance team, auditor, or partner bank will ask for, and most agent platforms simply don't produce it. The skills memory is the interesting long bet. Agents learn from resolved cases and propose updates to their own skills, and one governed memory spans support, payment ops, finance, and compliance. Teams running high volumes of near-identical cases get compounding value; teams running a handful of bespoke investigations will get less out of it. Where we'd push back: the two halves of the product don't tell the same story. The homepage is payment ops. The pricing page is coding agents — Claude Code, Codex, OpenCode, Gemini, GitHub Copilot, environment templates, BYOK with your Anthropic or OpenAI keys. Engineering teams can absolutely use it that way, but if you buy for payment ops and expect the coding-agent workflow (or vice versa) to be the default onboarding path, read the docs first. It is one harness with two
Researching Runtm? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Runtm actually fits — and what changes day-one when you adopt it.
A $48K vendor payout is declined. The ops lead triggers the bank-ops agent from Slack; the agent parses the bank's PDF, queries the ledger for the affected account, checks the KYC file and scans 90 days for the same pattern, then drafts a reply and attaches the evidence file.
Outcome: The return of funds sits in the approval queue until a human signs off on the $12,400 movement, and the run is stored with every step, query and cost for compliance.
They point Runtime at a monorepo, let it install toolchains and internal CLIs, and snapshot the result — the next boot takes 2.1 seconds. Team members launch Claude Code or Codex sessions from Slack with shared secrets and a per-agent spend limit.
Outcome: Engineers swap between coding agents without rebuilding tooling, and the team activity dashboard shows who ran what and what it cost.
A customer reports a duplicate charge. The agent checks payment events, confirms only one charge settled, attaches the ledger entry and the pending hold, and drafts a customer reply — but any refund above the configured threshold waits for approval.
Outcome: Replies go out faster with the payment breakdown attached, while money movement still requires a named approver.
Use Cases
- Spin up a Claude Code session with full project context and monitor its work in real time
- Automate incident triage by deploying a specialized agent that investigates alerts and posts findings to Slack
- Investigate declined vendor payouts and trace which rule or limit caused the failure
- Handle partner-bank RFIs, hold-harmless requests and returns of funds with ledger verification
- Prepare merchant reviews and KYB/KYC follow-ups with the ownership documents checked against the application
- Set company-wide spend limits and approval gates to control agent usage across teams
- Run background agents that auto-pause after completing tasks like updating dependencies
- Let finance query data in plain English via a private Slack agent with row-level scopes
Models Under the Hood
as of 2026-09-23
Limitations
- The free tier caps you at 500 credits, one user, one parallel session and a 4 vCPU / 4 GB sandbox, which is enough to evaluate but not to run a team.
- Teams starts from $99 per seat per month and lifts sessions and users to unlimited with 8 vCPU / 8 GB sandboxes, but the security controls buyers most often need — SSO (SAML), SCIM, audit logs, per-agent network allowlists, custom agent configurations and the SLA — are Enterprise-only.
- Self-hosting and bring-your-own-sandbox-provider are likewise Enterprise-only, so genuinely air-gapped or fully offline deployment isn't available below that tier.
- Shared templates and secrets, the team activity dashboard, and role-based access control arrive at Teams.
- Annual-only discounts and enterprise contract minimums are not published.
- The homepage's financial-operations framing also understates the coding-agent product, which can make scoping a trial confusing.
as of 2026-09-14
Verification history
We have re-verified Runtm 8 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 8 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Runtm tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free
$0/month
Ideal for
An individual developer or ops analyst evaluating whether sandboxed agents with approval gates fit their workflow.
What this tier adds
Free entry point: 500 credits, no credit card, 1 session, Claude Code/Codex/OpenCode only, 4 vCPU / 4 GB sandbox, community support.
Teams
From $99/seat/month
Ideal for
A product or payments-ops team of several people running agents against production-adjacent systems who need shared context and spend limits.
What this tier adds
Adds unlimited sessions and users, all coding agents, real-time collaboration, shared templates and secrets, spend controls, role-based access, team activity dashboard, 8 vCPU / 8 GB sandboxes, MCP and SDK/CLI/API access.
Enterprise
Custom
Ideal for
Regulated organizations — banks, fintechs, and their platform-security teams — that must self-host or run in their own VPC and prove compliance.
What this tier adds
Adds self-hosting on your own infrastructure, custom sandboxes up to 120 GB RAM, SSO (SAML), SCIM and audit logs, per-agent network allowlists, bring-your-own-sandbox-provider, SLA and a dedicated support engineer with a shared Slack channel.
Where the pricing makes sense
The company stage and team size where Runtm's pricing actually pencils out — and where peers do it cheaper.
Runtime's pricing suits funded engineering and fintech ops teams, not individuals. The $0 Free tier covers evaluation only (500 credits, 1 session, 4 vCPU / 4 GB). Teams at $99/seat/month undercuts assembling your own orchestration plus sandboxing plus audit logging, and is cheaper than most enterprise agent-governance platforms that quote custom-only. Enterprise is custom-priced and effectively bundles self-hosting, SSO/SCIM and an SLA — comparable governance products from established security
Setup time & first value
How long it actually takes to get something useful out of Runtm — broken out by persona, not the marketing-page minute.
Solo evaluation: minutes — sign up for the Free tier, no credit card, and a base 4 vCPU / 4 GB sandbox is available immediately; you can be in a Claude Code or Codex session on day one. Team rollout: hours to a day — you'll spend most of it importing the repo and letting Runtime build and snapshot your environment, then configuring scopes and approval rules per agent. Enterprise self-hosting:
Switching to or from Runtm
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From local agent CLIs (Claude Code, Codex, OpenCode) running on developer laptops: import your repo as an environment template and move runs into sandboxes with the same agents you already use.
- →From a self-built orchestration plus sandboxing stack: replace the custom scaffolding with Runtime environments, Mission Control visibility and the stored run record.
- →From manual payment-ops workflows in spreadsheets and shared inboxes: encode the procedure as an agent with ledger, email and Stripe tools and approval gates on money movement.
- →From GitHub Copilot-only workflows: keep Copilot if you like it — Runtime is agent-agnostic and lists Copilot alongside Claude Code, Codex and Gemini CLI.
- ↗To a single local coding agent CLI: you lose the governance layer and the audit trail, and would run agents against production systems with your own credentials.
- ↗To a general-purpose RPA or iPaaS platform: cheaper for simple deterministic workflows, but you'd rebuild the agent reasoning, scoped data access and approval model yourself.
- ↗To a fully managed no-code automation product: faster for non-technical users, but you give up BYOK, self-hosting and the stored run record.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Runtm”, and we withheld 6: 6 could not be judged, because “Runtm” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Runtm.
Official links
Tools that pair well with Runtm
Common stack mates teams adopt alongside Runtm, with the specific reason each pairing earns its keep.
Daytona
Daytona runs untrusted, AI-generated code in isolated sandboxes that start in under 90ms.
E2B
E2B runs secure Linux sandboxes so AI agents can execute code, process data, and use tools safely
Fimo
An autonomous website platform that runs on your own Git repo, with scheduled AI agents that keep improving the site after launch.
Featured Head-to-Head Comparisons
Runtm vs Spider Cloud
Choose Spider Cloud if you need fast, low-cost web scraping and crawling for AI agents or RAG — its Rust engine and 99.9% success rate deliver reliable structured data. Choose Runtm if you run coding agents (Claude Code, Codex, etc.) and need secure, isolated sandboxes with monitoring and governance. They solve completely different problems: data ingestion vs. agent execution environments.
Runtm vs Temporal Ai
Choose Temporal AI if you need reliable, durable orchestration for AI agents and long-running workflows that survive failures. Choose Runtm if your priority is providing secure, fast, pre-warmed sandboxes where coding agents can develop and deploy without infrastructure overhead. Both are open-source but serve fundamentally different needs.
Runtm vs Voyage Ai
Choose Voyage AI if your priority is high-accuracy retrieval in RAG pipelines, especially for specialized domains like finance or law. Choose Runtm if you need a secure, auditable sandbox for running multiple AI coding agents with governance and collaboration features. They serve fundamentally different needs—Voyage is about embedding quality, Runtm is about agent infrastructure.
Replit Agent vs Runtm
For teams that need governed, multi-agent infrastructure with observability and self-hosting, Runtm is the clear choice. If you want to prototype and deploy full-stack apps from one prompt with minimal setup, Replit Agent is faster and more approachable. Choose based on whether your priority is control (Runtm) or speed (Replit Agent).
Alternatives to Runtm
View all
Frequently Asked Questions
Best-of guides
Used Runtm? Help shape our editorial sentiment research.