AI-driven actions inside the systems your client's team already opens every day — HubSpot, Slack, CRMs, helpdesks, custom internal tools — with auth, rate limits, least-privilege access, and the usage dashboards the buyer needs to see ROI. Not a parallel chat surface that nobody opens.
Integration pod, predictable spine
These are the conversations agency owners describe when an integration brief is in front of them and the in-house team has shipped a side panel or a Slack bot but not yet AI that lives where the buyer's team already works.
The standalone chat tab worked in the demo, has a stand-up engagement chart that points down, and the buyer is asking why. We move the AI to where the work already happens — inside HubSpot cards, Slack message shortcuts, CRM record actions — and the usage dashboard finally goes the right direction.
Tokens expiring, scope errors at the worst moments, third-party rate limits hit in the middle of a busy hour, refresh-token races. We rebuild the integration with a real auth contract, scope minimisation, and a backoff and retry harness designed for production volume — not a single demo run.
Each feature was built by a different sprint with its own auth, its own audit gap, its own way of calling the model. We rationalise the surface: one shared auth and audit layer, one model-routing pattern, per-feature cost ceilings, and a single observability story across every surface.
Activations, weekly active users, actions taken inside the integration, time-to-first-value — none of it is being captured. We retrofit the measurement layer the renewal call needs, and put it on a dashboard the buyer can open without asking for a deck.
Integration work runs auth-first. The pod doesn't ship an AI surface into a real tool without an auth contract, a rate-limit harness, and an audit layer behind it.
Working session with your delivery lead and the buyer-side stakeholder. We map the surfaces (which tool, which views, which actions), the auth model, the rate-limit and tenancy constraints, and the ROI definition. NDA and SOW signed under Salt Technologies, Inc.
Architecture readout: AI placement per surface, OAuth scopes (with explicit minimisation), token-refresh and rate-limit strategy, audit and tenancy model, observability and ROI dashboard plan. The auth contract is signed off before any production credentials are issued.
Surface-by-surface build with weekly working review. OAuth flows wired with refresh-token races handled, rate-limit retries and backoff harness in place, audit logs landing on every action, and per-feature cost ceilings on AI nodes. Integration tests against the third-party sandbox run in CI.
Gradual rollout against real users, marketplace listing prep when the integration is going public, and the ROI dashboard live for the buyer with usage, activations, and time-to-first-value. Post-launch transitions cleanly into a Support & Maintenance retainer for third-party API changes, model upgrades, and v1.x feature work.
Integration work is long-arc once a real third-party surface is involved. The two shapes below are the engagements where production-grade AI integrations actually land — a new service line under your brand, or a delivery pod inside an active client SOW.
Capability lead + delivery support
Standing AI Integrations up as a defensible new service line on your rate card — capability lead, pricing template, monthly retainer with quarterly review. The most common shape when an agency is positioning as a HubSpot, Salesforce, or Slack AI partner.
Integration pod inside your SOW
Active client work that has been sold and now needs end-to-end integration throughput inside your shared channel — long-arc, structured, named delivery owner inside a buyer-side product surface.
Starting weekly rate for a single-capability AI pod. Multi-track AI pods (Integrations + Agents, Integrations + RAG) and pods that mix capabilities are quoted at the highest applicable rate. Final SOW is scoped against the brief.
We work inside the integration surfaces your team and your client already use — no parallel platform, no "we'll just rebuild it our way" surprise.
AI Integrations shows up most often where the agency already lives inside one or two platforms and the next brief asks for AI features that ship inside those surfaces — not standalone.
HubSpot partners shipping AI inside HubSpot — custom cards, workflow actions, contact-page UI, marketplace listings — where the standard automation surface stops and a real embedded AI feature begins. The single most common entry point for this capability.
Boutique design and product studios shipping AI inside custom internal tools, Slack apps, and bespoke admin surfaces — where the AI lives inside the existing UI the buyer's team already opens, not in a side tab nobody clicks.
Shopify partners adding AI inside merchant tools, the admin, and customer-side surfaces — where the integration sits inside Shopify itself or inside the helpdesk the merchant already uses, not a second app to install.
Every AI Integrations engagement runs on the same operating spine that protects long-arc retainers and Dedicated Partner Pods — contracted through Salt Technologies, Inc.
We don't email your client, join their calls, or appear in the proposal — unless you explicitly white-list a named engineer in the SOW.
We work in your GitHub, your model-provider accounts, your hosting, and your shared channel under aliases that fit your team's naming.
Mutual non-solicitation written into every MSA, with a defined window after the engagement ends. Same clause across every track.
MSA, NDA, and engagement SOW are issued by Salt — the Delaware C-Corp behind AgencyRelay.
The same operating spine sits underneath every AgencyRelay capability. Read the no-poach and confidentiality page for the contractual instruments behind these defaults.
Most agency owners deciding between an integrations pod, a single platform-engineer hire, or a freelance HubSpot / Salesforce developer weigh the same trade-offs. These pages address them head-on.
All four AI capabilities side by side — useful when the brief sits between Integrations and one of the other tracks and you want to see the routing decision laid out.
The most common shape for adding AI Integrations to your rate card — capability lead, pricing template, quarterly review on whether the new line earns its place.
When an integrations pod is cheaper, faster, and lower-risk than the next senior platform hire — and when staffing up is the right call.
Direct answers to the questions that come up on almost every AI Integrations scoping call.
Integrations put AI features inside an existing tool — a HubSpot card, a Slack shortcut, a CRM action — where the surface is fixed and the AI is one new capability inside it. Agents are the opposite — the AI decides which tool to call next, in what order, with state across multiple steps. If the buyer's team needs AI inside a surface they already open, it's an integration. If the buyer needs the model to reason across tools and take open-ended actions, it's an agent. Most briefs pick one cleanly; we route the ones that don't in week one.
The platform changes the build, not the engagement shape. HubSpot UI extensions, Slack apps with Block Kit, Salesforce Lightning Web Components, Zendesk apps, and custom internal tools each have their own SDK, review process, and rate-limit ceiling — and we ship into all of them. The architecture readout in week one names the platform-specific decisions (auth model, surface map, marketplace path) with the trade-offs written down.
OAuth 2.0 with explicit scope minimisation — every requested scope has a written justification in the auth contract. Refresh-token race conditions are handled at the connector layer, not at every call site. Rate limits get a backoff and retry harness, with a per-tenant ceiling so one heavy user can't starve the others. All of it ships behind audit logs the buyer's security team can read without asking us for help.
When the integration is going public, marketplace listing prep is part of scope — listing copy, screenshots and short video assets, scope and security review, the data-handling write-up the marketplace requires, and the back-and-forth with the platform's review team. When the integration stays private (a single-tenant build for one buyer), the marketplace step is skipped and the SOW reflects that.
Three layers, picked in week one with the buyer-side stakeholder. Activation — first useful action taken inside the integration. Engagement — weekly active users on the AI surface and actions per active user. Outcome — the buyer-defined metric (hours saved, deflection rate, conversion lift, time-to-first-value). All three land on a single dashboard the buyer can open without a deck — and the renewal conversation runs off that dashboard, not off our claim.
Production integrations into a third-party surface aren't a one-week capability — OAuth, rate-limit, audit, and review-process work add up. The most common starting shape is a 4–6 week pod inside an Invisible Delivery Team SOW, sized around the surface count and the marketplace path. For shorter integration work (a single Slack shortcut, a HubSpot workflow action, an audit-log retrofit), we'll quote a tighter window against the same weekly rate.
The starting weekly rate for a single-capability AI pod is $3,200 per week. Multi-track AI pods (Integrations + Agents, Integrations + RAG) and pods that mix capabilities (Integrations + Backend, Integrations + UI/UX) are quoted at the highest applicable rate. Final SOW is scoped against the brief; the rate is the floor, not a ceiling.
Integrations live or die in month two. Most graduate cleanly into a Support & Maintenance retainer post-launch — third-party API deprecations (the constant), OAuth scope changes (surprise emails from the platform), model upgrades on the AI nodes, marketplace re-review when the listing changes, and v1.x feature work inside a monthly envelope. Either the same pod or a smaller maintenance crew carries it on the same MSA, no second sales cycle.
Yes. IP ownership and assignment on delivered code, OAuth scaffolding, audit and dashboard layers, marketplace assets, and supporting artefacts is written into the MSA — the work belongs to your agency (and onwards to your client per your own client contract) on payment of the relevant invoice. The Salt Technologies templates are counsel-reviewed and shared before signing.
An AI feature inside HubSpot, a Slack app for an internal team, a CRM action surface, or a multi-tool integration. Either way, the conversation starts with the work — not with a deck.
