Every agency leader has opinions about tools — usually formed the hard way, through clunky migrations, abandoned platforms, and subscriptions that never earned their keep. So we asked agency founders, operators, and consultants a more useful question: if you were starting an agency from scratch today, what would you actually build around?
Their answers converged on a few consistent themes: Nobody wants a single tool that does everything and almost everyone wants a lean core with agency-specific fit for use cases like budgeting, retainers, and capacity planning. And increasingly, AI is being treated not as another app in the stack, but as the connective tissue holding the whole thing together.
Start with the hub: a PM tool or CRM at the center
The first decision isn't which tools to buy — it's what sits at the center. Mat Bennett, Agency Advisor, says the answer depends on the kind of agency you're building. "Some agencies seem best suited to have CRM at the centre of their systems, for others it is the PM tool," he explains. For relationship-driven agencies, his CRM pick is Attio, which he describes as "well-structured, fast, connectable and easily extendable with no pricing surprises." For agencies where project management is the key piece, he's watching Plane.so, noting that "it's API-first, has a clear data model, and a 'connect don't colonise' approach. Those are all hallmarks of next-gen systems to me." His guiding principle applies either way: "I don't need one tool to do everything. I need each tool to just do its own thing brilliantly."
I don’t need one tool to do everything. I need each tool to just do its own thing brilliantly.
If the hub is a PM tool, Kayla Keizer, Project Manager at Northern, makes the case for Productive.io, which she says is "specifically designed for agencies and agency work." In her view, it offers the folder structure of ClickUp without requiring the setup, while natively handling the things agencies actually need — budgeting, multiple contract types like retainers versus time-and-materials, invoicing, and capacity forecasting — all in one place. ClickUp is her runner-up, with the caveat that it demands far more manual configuration.
For smaller agencies, Marissa Taffer, Founder and President of M. Taffer Consulting, frequently recommends Asana. It has a shallow learning curve, and she finds it provides the "right amount of constraint" — enough structure to keep teams organized without letting them over-engineer their boards.
Keep the supporting stack lean
Once the hub is chosen, the strongest advice from operators is restraint. Rakia Finley, Founder and Managing Partner of Copper & Vine Studio, built her firm's core delivery system in-house on a custom MongoDB-based architecture after concluding that "off-the-shelf project management tools were not built for how we work, so we stopped trying to fit our delivery model into them." From there, everything else stays intentionally minimal: "Lean is the goal. Sustainable and useful is the standard." Google Workspace is the one tool she'd never give up, "because removing it would break the connective tissue between every other part of how we work." Her takeaway for anyone building a stack: "own what is core to how you deliver, then let the supporting tools stay simple."
Lean is the goal. Sustainable and useful is the standard.
Usama Moin, CEO and Co-founder of Bitrupt, takes a similar view from the vantage point of a 30-person agency spread across the US, Pakistan, and Germany. Internally, his pick is Notion: "it's flexible enough to bend to whatever we need, and that flexibility matters a lot in an agency because every client project ends up needing a slightly different structure." Clockify handles time tracking so the team can keep billable hours honest, and, like Finley, his team ended up building its own resourcing and reporting system, because "most PM or reporting tools are built for a single product team, not for juggling 15-20 client engagements at once with different scopes, currencies, and delivery timelines."
Notion is flexible enough to bend to whatever we need, and that flexibility matters a lot in an agency because every client project ends up needing a slightly different structure.
Taffer offers the scrappiest proof that lean works: a highly effective, fast-turnaround agency she worked with abandoned traditional PM software entirely and ran its whole banner-build operation on nothing but Slack and Airtable, used as a "souped-up spreadsheet."
Choose tools your clients will actually use
A stack that works beautifully inside the agency can still fail at the client boundary. Alexa Alfonso, Sr. Account Executive at Caylent, argues that for an agency, "whatever tool your clients will use is the best tool." Agencies can save themselves considerable friction, she notes, by simply adopting the client's software rather than forcing clients into the agency's stack. Her own ideal core includes Google Workspace, Slack, Miro, and Asana — the last of which she credits with transforming operations at an agency where she previously worked.
Whatever tool your clients will use is the best tool.
Moin's agency operates the same way in practice. Many of his clients already have their own tools, so his team simply works inside whatever the client uses and reserves its own stack for internal coordination.
The client-facing side of the stack can also do accountability work. Julia Rajic, Chief Operating Officer of Point Blank, is currently using Asana to build timelines that can safely be shared with clients — holding them accountable for approvals rather than letting deadlines slip in silence.
The tools (and traps) to avoid
The contributors were just as forthcoming about what they'd leave out of a fresh stack. Rajic warns against Resource Guru for agency capacity planning, having found it expensive and incredibly difficult to administer across multiple project managers; she moved her team to Float, which handled the drag-and-drop resource assignment they needed. Her sharper caution is about over-structure itself: at one agency, Wrike was fully integrated for everything, but its rigid checklists and templates ended up stunting the agency's creative culture — teaching her that over-structured tools can prevent "outside the box" thinking.
Taffer's warning list starts with Monday.com, where she finds the learning curve too steep and the extreme customizability a liability: teams over-engineer their processes and turn the platform into a "junk drawer." She also recalls an agency running Workamajig where the system was so clunky the team built schedules in TeamGantt and manually copied the data back over — a double-entry tax no new agency should sign up for.
Keizer's pick to avoid is Basecamp, which she finds too client-facing on its own, always requiring a secondary tool to actually manage the project.
AI is becoming the connective layer
The newest shift in agency stacks isn't another platform — it's what sits between the platforms. Bennett has coined a term for it: "Claudification," which he describes as "a slightly tongue in cheek term I use to describe this pattern of people moving capability away from traditional SAAS offerings (particularly the monolith 'do everything' services) and into Claude AI." The result, he says, is that "this approach is giving agencies highly customised stacks for tiny outlays. Systems get shaped overnight to accelerate their specific workflows, rather than workflows being shaped by the limits of the tools available."
He's even seeing "disposable" software spun up to support specific projects, then scrapped when the project is complete. Crucially, he argues the driver isn't usually cost savings — "it is more frustration with features that are built for everyone not meeting the individual needs of a business." AI also smooths cross-system integration, because it can "reason between the gaps" where traditional integrations broke on messy, non-standardized data.
Kazim Qazi, Co-founder of TechnBrains, has embedded AI directly into delivery operations. Atlassian Intelligence handles first-pass ticket triage in Jira: "it reads a raw request, drafts a scoped breakdown, flags missing acceptance criteria, and routes it to the right engineer before a human touches it." He calls AI-assisted triage the tool most agencies underuse: "Most agency backlogs die from vague scope, not lack of tools, automating that first triage pass is what actually moves a stuck sprint, more than any new PM platform does." On the quality side, Claude Code Review runs automatically on every pull request — "it reviews the whole repo, not just the diff, which catches cross-file issues before a client-facing bug ever ships."
Most agency backlogs die from vague scope, not lack of tools, automating that first triage pass is what actually moves a stuck sprint, more than any new PM platform does.
For smaller teams, the AI layer can be simpler. Jennifer Goebel, Project Coordinator at Baker Marketing Laboratory, says her small healthcare marketing agency leans on Claude for copywriting because it captures a natural tone of voice much faster than ChatGPT, and uses NotebookLM to store all monthly client reports and background information so anyone on the team can query past client data when writing new content.
Vladimir Krstić, Founder and CEO of EVI Solutions, adds a governance point: rather than letting AI use fragment across the team, his agency runs Intrascope internally because "it gives the team one shared workspace for multiple LLMs, reusable prompts, shared context, and better visibility instead of scattered personal AI accounts." The standard he holds any tool to is the same: "the tools should make good delivery easier to repeat. They should reduce context loss, clarify responsibilities, support review steps, and help the team move from client request to finished delivery without important information getting buried across chats, documents, or personal accounts."
Build the stack around how you deliver
Strip away the specific product names, and the advice from these leaders forms a clear blueprint. Pick one hub — PM tool or CRM — that matches how your agency actually creates value, and demand agency-native features from it rather than bending product-team software to fit. Keep everything around that hub lean, and be willing to build the one piece that's genuinely core to your delivery model. Let clients pull you toward their tools when it reduces friction, and steer clear of platforms whose flexibility invites over-engineering. Then let AI do what no integration ever could: reason across the gaps, so the stack works as one system instead of a dozen subscriptions.
Want more insights like these? Sign up for a free DPM account to hear from more experts like these.
