ChannelGate is the company-governed AI operating system.
Bring your team, Claude Code and Codex together with permissions, memory, connections, approvals and shared work in one governed environment. ChannelGate coordinates the work in Slack, Microsoft Teams (Beta) or Google Chat (Beta).
The multiplayer advantage
Using ChatGPT, Claude, or Codex privately makes one person faster.
A governed AI agent in the channel makes the whole company more capable.
The work, decisions, and context stay visible in the thread. AI champions lead by example, and every teammate builds confidence through real work.
AI OS for the whole company
From a weekly sales check to a complete product catalog: see what an AI agent can do with your tools, context and reusable skills.
The agent checks open HubSpot deals against emails, meetings and tasks, then shows each rep who needs a reply, a follow-up, reactivation or a decision. Each nudge includes the relationship context and next step, so the team can act from the same conversation instead of rebuilding the history.
HubSpot · Gmail · meetings · tasks · opportunity-audit skillThe agent searches HubSpot for relevant prospects and past customers, enriches their profiles, and checks fit, existing relationships and recent contact activity. It then adds the selected contacts to your outreach platform, ready for an automated event invitation campaign.
HubSpot · LinkedIn/company enrichment · outreach platformThe agent reads the call and follow-up context, separates the work into scopes, and develops budget options using your proposal guidance. Your presentation skill turns this into customer-ready slides with delivery phases, assumptions and next steps.
Call transcript · proposal guidance · presentation skillThe agent reviews bank transactions, separates internal transfers from operating spend, and finds unusual or misclassified costs. It explains what drove the change and links the supporting records, so the team can distinguish recurring costs from one-off spending and focus its investigation.
Bank records · KPI analysis · transaction classificationThe agent compares subscription fields with contracts, invoices and database records, applying the reconciliation rules your team has agreed. It records the evidence behind each match or discrepancy and produces an exception list, so reviewers can focus on ambiguous dates, missing documents and conflicting values instead of checking every field manually.
Contracts · invoices · database records · reconciliation rulesThe agent gathers emails, meeting discussions, commitments and documents into a handover dossier. It organizes the supporting files and adds the handover reference to the CRM, with missing source material called out. The delivery team starts with the agreed scope, open questions and relationship history together, ready for the kickoff.
Gmail · meeting transcripts · Google Drive/Docs · HubSpot · onboarding skillThe agent follows your presentation skill to develop the story, slide structure and visual style from a brief or existing material. Your team can refine content and design in the same thread. The agent checks the rendered layout, corrects clipping or spacing, and publishes a shareable deck that can be updated as the discussion evolves.
Presentation skill · brand assets · browser checks · shared deckThe agent turns customer requirements and source documents into a clickable prototype with simulated demo data, different user views and the key workflow steps. It tests the demo journeys and shares a live preview, so your team and customer can explore the experience, identify missing requirements and refine the scope before production implementation.
Requirements · prototype · browser journey tests · hosted previewThe agent checks run histories across connected projects, reads the failure details and traces the error to its cause. It distinguishes code defects from expired authentication or missing inputs, then reports the practical next step. Where the work is authorized and the necessary access is available, it can make a repair and verify the result.
Automation run histories · error traces · code fixes · deployment checksThe agent collects product details, images and documents, enriches and translates catalog fields, and applies your categorization rules. It prepares export files for products that are ready, while keeping missing or discontinued items visible as exceptions. Your team can review the records, refine the rules and continue the next batch from the same conversation.
Supplier sources · product records · translation · assets · export filesHow it works
One conversation. Six visible steps. Follow Ana’s request from her inbox to an approved reply.
To: Makeitfuture · “Friday works for the renewal call. Thank you!”
The reply confirms Friday for the renewal call.
A colleague mentions the AI agent where the team already works. No new app to open or special prompt format to learn. Everyone who should see the request can follow it.
Before work begins, ChannelGate verifies the requester, the channel’s mode—Read-only or Worker, with Auto configured separately—and any required approvals. Nothing starts until every check passes.
The AI agent works inside a private, isolated environment containing only that channel’s instructions, skills, files, memory, tools, and approved connections. Everything else remains out of reach.
Claude Code or Codex performs the work using only authorized tools and connections. Channel connections belong to the channel; personal connections remain bound to the person who linked them.
For actions configured to require approval—such as sending email, spending money, or changing settings—the AI agent pauses for the designated approver. The decision—and who made it—stays in the thread.
The outcome appears in the channel with the work performed, approvals, time, cost, and what the channel learned. The next request begins with shared context.
EVERY CHANNEL GETS ITS OWN AI AGENT
A channel is more than a chat with a bot. It gives an AI agent a defined role—with its own instructions, skills, approved connections, memory, automations, and approval rules—in its own container. Worker channels keep their scope; admin channels can share the gateway home when enabled. Choose a channel to see what it can access.
Nothing from other channels is visible in this workspace.
Qualify opportunities, surface stalled deals, and follow pricing guardrails.
Objections, decisions, stalled deals, and approved pricing exceptions.
Nothing from other channels is visible in this workspace.
Follow payment terms, approval limits, and the exception policy.
Vendor terms, recurring anomalies, and approvals.
Available today · Gateway-home sharing enabled
With gateway-home sharing enabled, the operator’s home is mounted inside this admin channel—including other channel folders, project files and saved conversations. Admin turns can inspect and edit those files.
Review work across channels, inspect saved conversations, update project files, and resolve blockers.
Priorities, leadership decisions, dependencies, and unresolved blockers.
Each channel has its own container. Administrators control which files and connections it can access.
Saved secret values cannot be displayed again. Authorized work receives the credentials it needs at runtime.
Personal connections remain bound to the person who linked them. Channel connections are governed separately.
Choose the tools and approval rules for each channel. Each channel’s permissions remain separate from its connected identities.
BUILT ABOVE THE AGENTS
ChannelGate is the company-governed AI operating system. It coordinates your team and leading agents with identity, permissions, approvals, connections, memory and cost tracking. Choose the best agent for each job—and adopt better ones as they emerge—without rebuilding how your company works.
Where your team talks
Who does the work
Anthropic
Choose the AI agent you prefer for this channel or conversation. ChannelGate AI OS keeps the surrounding experience consistent whichever agent you select.
Set a default per channel. Change it for any conversation. The AI agent can change; the governed context stays.
One consistent experience
Optional fallback keeps work moving.
If the selected AI agent is unavailable, ChannelGate AI OS can continue with another agent approved for that channel—preserving the conversation, workspace, permissions, approvals, memory, and history.
Only approved agents may be used, and every switch is recorded in the thread. A conversation with a manually pinned AI agent never switches automatically. These controls illustrate configuration; they do not change your workspace.
Memory with boundaries
Each channel keeps an editable, reviewable memory of decisions, preferences, and operating rules. Corrections are saved with their source and scope, so future work begins with agreed context—not hidden assumptions.
Always available, still governed
Start a task from your phone, continue on your laptop, or close both. ChannelGate runs in your environment—not on one person’s device—so long-running work, schedules, and monitors keep running and report progress back to the channel. The AI agent still pauses whenever approval or human attention is required.
Analyze the site, identify SEO improvements, update page titles, descriptions and internal links, then check the changes and report the results back to the channel.
Recurring optimization loopCollect the previous week’s revenue, compare it with the week before, explain the main changes and post the report in the channel.
Weekly reportCheck the CRM and invoicing system, identify stalled deals and overdue payments, and prepare the next follow-ups for the team.
Daily follow-up checkEU AI ACT
Make responsible AI use part of how your team works. ChannelGate brings practical controls into the conversation, helping people follow procedures, review decisions and stay accountable.
Configure actions to wait for human approval and keep the approver’s decision in the thread. Your organization chooses who reviews the work and when intervention is needed.
Clear accountabilityRequests, results and approval decisions stay together in the conversation. Teams can review the context behind the work and decide what additional records their use case requires.
Context for reviewersChannel instructions, reusable skills and configured tool access help teams work consistently with approved connections. Operators set the access policy; instructions guide the agent’s work.
Consistent proceduresAI champions demonstrate useful workflows and review real examples with teammates. This supports onboarding and AI literacy alongside the training and policies your organization provides.
Informed adoptionEnforced organization instructions and mandatory organization skills are available Enterprise controls that support consistent policy and procedure application. They guide and standardize AI work; they do not guarantee model behavior or compliance.
Enterprise controlSchedule an admin-enabled AI agent to review work across your other agents, surface blockers and flag improvements for your team. Its reach depends on the configured admin workspace access; people remain accountable, and this review does not replace required human oversight or authorize use of another person’s personal connections.
Admin-enabled reviewThe AI Act takes a risk-based approach. High-risk deployer duties and transparency obligations have their own scope; your organization needs to assess which rules apply.
Compliance depends on the use case, role, configuration and organizational processes. ChannelGate supplies supporting controls.
Easy to configure
Set up a channel agent the way you’d onboard a new colleague: give it clear instructions, the right skills and connections, then refine it through real work.
Created and implemented by Makeitfuture
Makeitfuture helps you deploy ChannelGate in your environment, design governed AI agents around real work, turn company knowledge into reusable skills, and equip internal champions to keep improving them after launch.
Built and implemented by Makeitfuture — Make AI Partner of the Year
Start open. Scale deliberately.
Self-hosted evaluation
No registration required
Deploy from GitHubFree organization license
Commercial agreement
Infrastructure and AI-provider usage are billed separately. Cost-tracking history limits apply only to cost records, not AI usage allowances or general conversation retention.
Implementation by Makeitfuture
Optional one-time service for deployment, guided build sessions and handover. Your organization license stays free. Review the 30-hour scope.
THE FUTURE OF TEAMWORK
Your best agents. Your whole team. One conversation.