A dedicated tech pod that builds custom AI for your business — and the platform to run it
Think of Retanu as OpenRouter for your whole organization — one platform across every AI provider. The difference: the Black Gibbon Tech Pod builds the custom AI on top, so you get the outcome without wiring up LLMs and infrastructure yourself.
One pattern, every vertical
If your organization has sub-units that each need custom AI — and you'd rather a dedicated tech pod design, build, and run it than staff an AI team — Retanu fits. A “workspace” is whatever your sub-unit is.
SaaS platforms
Per-customer routing and spend caps for inference features, without building a gateway.
Professional services
Firms that already bill by client, matter, or engagement get per-unit cost attribution that fits their existing habit.
Healthcare orgs
MSOs, DSOs, hospital systems. Regulated, multi-unit, and used to outsourcing ops.
Multi-location operators
Restaurant groups, retail chains, dealerships, hotel groups. Each location is a workspace with its own usage line.
Financial services
RIA networks, banks, MGAs. Regulated, multi-unit, native chargeback.
Education & government
Districts, universities, city agencies. Budget-attribution-driven, often FERPA-regulated.
Membership bodies
Associations, co-ops, buying groups. Managed inference ops as a member benefit, metered per member.
Agencies & consultancies
Digital agencies, IT consultancies, BPOs. Each client or project gets isolated inference ops with its own cost line.
The problem
- A different dashboard and API key for every inference provider
- No idea what each unit is spending until the bill arrives
- Someone has to pick the right model for every use case
- No budget limits — a runaway app can blow your spend
- Attributing cost to the right client, team, or location means spreadsheets
- One place for all providers and all workspaces
- Real-time spend visibility, per workspace, per model
- Automatic routing picks the best-value model for the task
- Spend caps and rate limits that actually block, not just alert
- Usage reports generated automatically — per client, location, or team
What you get
Smart routing
Every request is classified and sent to the best-value model automatically. Simple tasks go to cheaper models; complex tasks go to stronger ones. If a provider is down, it fails over to the next one.
Spend controls
Set a monthly budget per workspace. When it's hit, requests are blocked — not just logged. Rate limits prevent runaway usage. You stay in control.
Per-unit cost attribution
Every request is metered and attributed to its workspace. Generate usage reports per client, location, or team — with your own rates applied. Fits naturally into existing chargeback or invoicing workflows.
Drop-in API
Works with any app that speaks OpenAI's format — which is almost everything. Change one URL and one key. No code rewrite, no vendor lock-in.
The platform and the team — without hiring for either
Most companies need two things they don't have: an inference ops platform to manage routing, budgets, and attribution — and an ops team to run it. Retanu is the platform. Black Gibbon Tech Pod is the team. You get both — or either one.
Self-service
You have an ops person or team. Sign in, connect your provider keys, and manage everything through the console. The platform handles the hard parts.
Assisted setup
We configure your Retanu instance, set up workspaces and routing, train your team, and hand it over. You run it from there.
Fully managed
Black Gibbon Tech Pod handles setup, monitoring, routing changes, and reporting. Your apps call the API; we keep it running. No ops hire needed.
Case studies
The work we run. Every engagement is the same promise: we build the workflow, then we operate it — your team gets the outcome without hiring an AI team or running infrastructure.
Document Intelligence
We built retrieval and Q&A over their full regulatory corpus — every answer cites its source page.
We run the corpus: new regulations indexed as published, accuracy reviewed continuously.
Conversational Data Access
We built plain-English access to their enterprise data — questions in; tables, charts, and explanations out. No SQL, no ticket queue.
We run the translation layer: query accuracy, new data sources, and who can ask what.
Customer-Service Agent
We built a brand-consistent agent across their storefront — live catalog search, policy answers, order help, clean handoff to humans.
We run the agent: reply quality reviewed by our human-QA team, knowledge kept current.
Intelligent Intake
We built automated intake — documents read and verified on upload, applications screened, exceptions flagged for human review.
We run the pipeline: extraction accuracy, screening rules, and the exception queue.
Social Monitoring & Auto-Response
We built always-on social listening with intent detection, on-brand automated replies, user risk-scoring, and a full audit trail.
We run the operation: reply tone QA'd against brand guidelines, risk rules updated, escalations routed.
Compliance & Trade Analytics
We built automated analysis of their customs data — tariff classifications checked at scale, discrepancies flagged, audit-defense documentation generated.
We run the reviews: rules updated as trade regulations change, findings queued for their experts.
Every engagement runs on Retanu — isolated, metered, and continuously optimized, so it gets cheaper and better without anyone on your team touching a model.
Results shown are from deployments delivered by our team and build partners; outcomes depend on your volume, data, and workflow. Clients anonymized.
How it works
Connect your inference providers
Sign in with Google. Paste your API keys for OpenAI, Anthropic, DeepInfra, or Google. You're live in under a minute.
Add your workspaces
Create a workspace for each client, team, or location. Set its budget, quality tier, and provider preferences. Each one gets its own endpoint and key.
Monitor & attribute
Watch spend in real time, per workspace and per model. Generate usage reports with your rates applied — ready for chargeback or invoicing.
How Retanu compares
| Capability | DIY | OpenRouter | Retanu |
|---|---|---|---|
| Automatic model selection | Build it | Model routing only | Built-in, 3 quality tiers |
| Budget enforcement | Build it | No | Spend caps and rate limits, fail-closed |
| Per-unit data isolation | Build it | No | Database-enforced, QA-tested |
| Per-unit cost attribution | Build it | No | Metered per workspace, reports with your rates |
| Managed ops team | Hire your own | No | Black Gibbon Tech Pod |
| Your keys, your pricing | Yes | 5% over 1M req | Your keys, zero surcharge |
Questions
What counts as a "workspace"?
Whatever your sub-unit is — a client, customer, department, practice, location, project, or member. You can even rename the label in Settings (e.g., "Matter" for law firms, "Location" for franchises).
Do I need an inference ops or engineering team?
No. Retanu is the platform — you can run it yourself through the console. If you don't have an ops team, Black Gibbon Tech Pod can be that team for you.
Who pays the inference providers?
You connect your own provider accounts. The provider bills you directly at their standard rates. Retanu doesn't add a surcharge. You decide how to price it for your own clients or teams.
Is data kept separate between workspaces?
Yes. Every organization's data is isolated at the database level with row-level security. One organization cannot see another's data, keys, or logs. Verified by 16 automated tests.
What if I already use OpenAI's API?
Retanu speaks the same format. Your app changes one URL and one API key — no code rewrite. You can always switch back.
What's Black Gibbon Tech Pod?
Your inference ops team — without the hire. We set up your Retanu instance, configure routing and budgets, monitor usage, and handle day-to-day operations. Available as assisted setup or fully managed.
Get the platform and the team
Sign up for early access. Run Retanu yourself, or let Black Gibbon Tech Pod handle the ops.