A grounded look at what’s changing (and what’s not) 

AI is no longer just a backend feature or a buzzy headline. It’s starting to shape how teams actually work together, in how they communicate, how they make decisions, and how they ship. But at Creed, we’re not treating it like magic. We treat it like any other tool: with curiosity, clarity, and care.

What’s surprised me most isn’t that AI can write a paragraph or suggest a line of code. It’s that AI is quietly becoming a collaboration layer for teams to compress knowledge, reduce friction, and stay aligned when the pace gets chaotic. We’re seeing that shift show up in our client work and inside our own delivery teams, and it’s changing the “shape” of a project in practical, measurable ways.

This post is for leaders and practitioners who want new collaboration techniques, not hype. It shares patterns we’re seeing in the wild across Slack and Teams, Jira and Confluence, knowledge bases, and engineering workflows, along with a few “steal-this” ways our development, design, and project management teams at Creed are using AI to collaborate better.

Collaboration with AI isn’t one thing. It’s four.

When clients tell us “we want to use AI,” what they often mean is: we want work to move without breaking people.

In practice, the best collaboration wins fall into four buckets:

1. Sensemaking: turning noise into a shared understanding (summaries, clustering, “what changed?”).
2. Drafting: getting from blank page to workable first version (tickets, specs, emails, test plans).
3. Decision support: clarifying options and tradeoffs through research (buyer and user personas, pain points, industry challenges, competitor approaches, assumption stress-testing).
4. Orchestration: routing tasks and keeping the machine moving (handoffs, checklists, follow-ups).

That last one is where AI starts to feel like a teammate, not a tool. It’s also where you need the most intentional design and governance.

What we’re seeing with clients: AI is moving into the tools teams already live in

Most organizations don’t want another dashboard. They want AI to show up where work already happens: messaging, project tracking, documentation, and dev environments.

Slack & Teams: AI as the “front door” to shared context

The most common collaboration pain we hear is some version of:

“I know we talked about this… but I can’t find it.”

That’s why Slack/Teams is becoming the natural home for AI copilots. One of the example patterns we’ve been building and testing is a Slack AI Assistant that pulls information across knowledge tools (like Notion or Confluence) and pushes summaries back into Slack, so the team gets answers without going on a scavenger hunt.

Practical ways we see teams using this pattern:

• Thread compression: summarize a long conversation into “Decisions / Open questions / Next steps.”
• Async standups: turn “what I did / what I’m doing / blockers” into a clean update for the channel.
• Rapid onboarding: ask the bot “What is this project? Where’s the spec? What’s the latest decision?”

If this sounds familiar, there’s a simple way to try this with your own team.

Copy and paste  this prompt:
“Summarize this thread for someone who missed it. Include: (1) decisions made, (2) open questions, (3) action items with owners if mentioned, and (4) any deadlines or risks.”

The best teams don’t use AI to replace conversation. They use it to make conversation retrievable.

Jira & Confluence: AI as the bridge between talk and tasks

The second most common pain:

“We had a great conversation… and then nothing happened.”

This is where AI shines as a translation layer—turning messy inputs into structured work. In our own internal thinking, the “quietly useful” uses are things like summarizing feedback before a sprint or drafting acceptance criteria faster—work that speeds up the flow without adding confusion.

High-impact use cases:

• Ticket grooming at speed: Convert a meeting transcript or notes into user stories + acceptance criteria.
• Scope clarity: Ask AI to identify ambiguity (“What’s missing to estimate this confidently?”).
• Sprint planning support: Cluster feedback themes and propose a draft prioritization.

Try this workflow:

1. Paste raw notes (or a Confluence page) into secure AI.
2. Ask for 3 candidate Jira tickets, each with: user story, acceptance criteria, and definition-of-done checklist.
3. Ask AI to flag unknowns that require human decisions.

This is especially powerful when paired with a human rule: AI can draft; humans decide. That principle keeps teams moving while protecting quality and accountability.

Knowledge management: “internal GPTs” that unify fragmented answers

Many companies already have the knowledge they need. The challenge is getting to it at the moment it matters, without breaking focus or relying on the one person who happens to know where everything lives.

That’s why teams are starting to experiment with internal GPTs and retrieval-based assistants that can pull answers from multiple systems, including tools like Notion, Confluence, Slack, and SharePoint. In these setups, AI is not creating new knowledge. It is helping teams access what already exists, across places where information is often fragmented.

In our AI advisory work, we often see this explored through proof-of-concept efforts around AI-enhanced knowledge management. Sometimes that means aggregating internal documents and messages so employees can ask questions and get consistent answers. Other times, the first step is simply giving teams secure access to approved AI models inside company boundaries, so they can use AI without risking data exposure.

The collaboration benefit is not just faster answers. It is fewer interruptions, fewer context switches, and less “tribal knowledge” locked in a single person’s head. Over time, that shared access to information makes collaboration calmer and more resilient.

AI governance: collaboration requires trust

As soon as AI starts summarizing, recommending, or routing work, trust becomes a product requirement. That’s why we consistently push clients (and ourselves) toward clarity: labels, permissions, audit trails, and an explicit understanding of what the AI is doing and why.

In our internal AI implementation planning, we’ve treated governance, training, and rollout as core work and not an afterthought. Because adoption dies quickly when teams feel unsure or exposed.

How we use AI at Creed for better collaboration

We’ve learned that AI collaboration isn’t “one tool.” It’s a set of team habits. 

Inside Creed, these collaboration shifts do not show up as one big initiative or a single AI tool. They show up differently depending on the work being done. Below are a few examples of how our design, project management, and engineering teams are using AI in small, repeatable ways to collaborate more effectively.

Design: moving from “control” to “co-work”

Designing for AI means designing for interaction. We’re seeing (and building) interfaces that include prompt fields, regenerate/edit paths, confidence indicators, and clear labeling. These are small UI decisions that keep humans in control while AI carries part of the load.

Internally, our designers use AI in a few consistent ways:

• Generating multiple copy directions or layout concepts, then converging with human taste.
• Summarizing research notes into themes, then validating against real evidence.
• Using clarity checks such as “What would confuse a new user here?” (this really is an underrated prompt.)

A subtle but important shift: AI makes it cheaper to explore. The design team’s job becomes “explore widely, decide deliberately.”

Project management: fewer meetings, better alignment

Project managers have always been the collaboration engine. AI doesn’t replace that. It gives PMs leverage.

A few patterns we’ve leaned into:

• Using AI to turn raw updates into crisp summaries so the team stays aligned without an hour-long status call.
• Drafting simple  “risk registers” from notes to flag dependencies that were implied but not stated.
• Clustering retro feedback into themes so retros lead to decisions rather than just venting.

These are the “quiet wins” that make teams feel calmer while delivery speeds up.

Development: from “coder” to “orchestrator”

On the engineering side, the most meaningful shift isn’t autocomplete—it’s agentic workflows.

We’ve been exploring tools and practices that let AI handle multi-step tasks while humans supervise, validate, and steer. In our internal material on Claude Code, a few collaboration-friendly practices stand out:

• Documenting context for the AI by using a CLAUDE.md or similar file to capture coding standards, directory structure, and architectural decisions so the AI follows team norms.
• Planning before executing by using a plan mode that outlines steps in advance and creates a natural pause point for human review.
• Working within this structured loop: Investigate → Plan → Implement → Test → Resolve. It’s predictable, reviewable, and team-friendly.
• Using small teams of specialist agents, such as architecture, testing, or refactoring, to parallelize work while keeping clear boundaries and merging only after review.

This improves collaboration because it changes what engineers bring to the team. Less time spent grinding through boilerplate, more time spent aligning on architecture, risk, and tradeoffs.

In other words: AI doesn’t just speed up code. It makes it easier for engineers to stay present in the team’s work.

8 collaboration techniques you can try next week

Here are practical techniques that don’t require a reorg or a moonshot platform migration:

1. The “missed it” summary (Slack/Teams): decisions, open questions, next steps.
2. The meeting-to-tickets bridge (Jira): raw notes → stories + acceptance criteria + unknowns.
3. The “draft-first” spec (Confluence): outline the doc, then collaborate on what matters.
4. The retro synthesizer: cluster feedback, propose 3 experiments for next sprint.
5. The knowledge concierge: “Where is the latest doc? What changed since last sprint?”
6. The code review buddy: AI drafts review notes; humans validate and decide.
7. The onboarding copilot: answer repeat questions from SOPs + quizzes (we’ve even explored this as a formal onboarding pattern).
8. The governance checklist: permissions, labeling, audit trails—trust is part of collaboration.

Final take: AI isn’t replacing teams. It’s reshaping how teams work

AI isn’t transforming everything overnight. But it is changing how teams design, build, and collaborate—especially when it’s embedded into real workflows instead of bolted on as a novelty.

At Creed, we’re focused on tools and patterns that support real work: staying in control, using time well, and building trust in the systems we ship.

And as we help clients evaluate readiness, build proof-of-concepts, and create rollout playbooks, we keep coming back to the same principle: AI creates value when it reduces friction and increases clarity.

If you’re experimenting right now, start small—but start where collaboration hurts most. Pick one workflow (threads → summaries, notes → tickets, docs → drafts), set guardrails, and measure whether your team feels more aligned a week later.

That’s the real benchmark: not “did we use AI,” but did we collaborate better because we did?