Nathan Weill here.

I'm the CEO @ Flow Digital, where we help companies unleash their full potential by strategically automating every inch of their workflows. Each week, I share hot AI and automation tips to help you move your business into the future successfully.

Pick the Right Proposal Platform for Your Sales Motion

Sales teams lose deals in the gap between pitch and payment. Static PDFs get buried in inboxes, pricing updates require manual back-and-forth, and signatures stall because contracts live in a separate tool. Two platforms close that gap in different ways.

PandaDoc is built for operationally complex, document-heavy organizations. It handles proposals, quotes, contracts, e-signatures, and payments in a single workflow.

Qwilr turns proposals into interactive, trackable web pages designed to stand out in competitive deals.

The quick rule: If your team needs end-to-end document ops across sales, legal, and HR with approval workflows and compliance controls, PandaDoc is the stronger fit. If your primary goal is visually differentiated proposals with buyer engagement tracking for high-touch B2B deals, Qwilr delivers faster.

Latest Updates to Start Using Now

HubSpot's Spring 2026 Spotlight rolled out Smart Deal Progression and other tools to allow teams to focus more time on sales and less time on CRM updates.

Key new features:

Three updates worth using right now:

Layer these together and HubSpot starts handling the admin work your reps were never meant to do.

Not sure which of these updates are worth turning on first? Schedule a Discovery Session and we'll map them to your sales goals.

1. Start with five questions, not fifty metrics.

Before building anything, write down the five questions your team asks every week. "Which channels are generating qualified leads under $50 CPL?" is useful. "Total impressions" is decoration.

If a metric doesn't answer a recurring business question, cut it. Luzmo's 2025 industry survey found that 37% of users say dashboard data isn't clear or actionable enough. The root cause is almost always scope creep: too many metrics, not enough focus.

2. Assign every dashboard an owner.

Unowned dashboards rot. Without a designated person responsible for accuracy and relevance, metrics go stale, definitions drift, and teams stop trusting the numbers.

One person per dashboard should review it quarterly and answer: Is this still accurate? Is anyone acting on this? Sigma Computing's guide to retiring dead KPIs recommends adding a "deprecated" label to unused dashboards and archiving them on a set schedule rather than letting them clutter your workspace.

3. Build role-specific views.

Your CMO and your paid media manager need different numbers. Forcing both into one dashboard means neither gets what they need.

One B2B SaaS team saw dashboard adoption jump from 23% to 87% within two weeks after splitting a single executive dashboard into three role-based views. The CMO saw monthly MRR impact from marketing. The paid media manager saw daily budget pacing and CPL. The content team saw organic traffic and engagement.

Same data. Different frames. Dramatically higher usage.

4. Build a shared metric glossary.

When "conversion" means something different to marketing than it does to sales, every cross-functional meeting becomes a debate over whose numbers are right instead of what to do about them.

Databox's research on data literacy gaps found that metric misalignment is the most common and most invisible barrier to data-driven decisions. Their recommendation: standardize definitions at the executive level, not the analyst level. If leadership doesn't ratify the glossary, teams won't adopt it.

A simple shared document defining what each metric means, where it comes from, and who owns it removes 80% of the friction in cross-functional reporting.

5. Schedule a monthly "so what?" review.

Dashboards decay the moment you stop paying attention to them. One analysis of dashboard rot describes abandoned dashboards as "organizational attention grave markers," marking the moment a team moved on.

Block 30 minutes each month to walk through your active dashboards. For each metric, ask one question: "What action did this drive in the last 30 days?"

If the answer is nothing for two consecutive months, archive the metric. You can always bring it back. The goal is a dashboard that reflects current priorities, not last quarter's assumptions.

6. Connect every metric to a specific decision.

This is the rule that ties the other five together. As researcher Dr. Victor Gonzalez noted, teams with dashboards but no interpretive structure end up producing slide decks instead of decisions. Without a clear link between a metric and the action it should trigger, data becomes background noise.

For every KPI on your dashboard, document:

When every metric has a decision, a threshold, and an owner, your team stops skimming dashboards and starts using them.

The bottom line: Better dashboards don't come from better software. They come from fewer metrics, clearer owners, and a direct line between every number on the screen and a decision someone needs to make. Teams that treat their dashboards as living tools rather than set-it-and-forget-it displays will always outperform those drowning in data they never act on.

Want help turning your existing dashboards into tools your team actually opens every morning? Let's talk about it.

Here's What Your Team Needs to Know

What Are Skills for AI, and How to Start Building Them

Your AI tools are only as good as the instructions you give them. Skills make those instructions portable, reusable, and consistent across your entire team.

The Problem: Every AI Session Starts from Zero

Your marketing lead writes a great prompt for generating client reports. It produces clean, on-brand output every time she uses it. Then a colleague tries the same task in a new chat window and gets something completely different.

This is the reality for most teams using AI in 2026. What used to be called "prompt engineering" is evolving into what the industry now calls context engineering: managing everything the model sees, not just the words you type.

The core problem is repeatability. A great prompt in one person's hands can become a messy output in another person's hands. Institutional knowledge about how to get good results lives in someone's clipboard, a Slack thread, or (worst case) only in their head.

Skills solve this by turning your best-performing workflows into packaged, portable instructions that any AI tool can follow automatically.

What Exactly Is a Skill?

In plain English: it is a text file that tells an AI agent how to do a specific task the same way, every time, for every team member.

Think of it like an SOP (Standard Operating Procedure) for AI. Industry experts have noted that this modularity makes AI development feel less like prompt engineering and more like onboarding a new employee with a clear set of standard operating procedures.

What makes Skills different from just saving a good prompt:

This Applies to More Than Engineering

The early adopters were developers, but the format is built for anyone who can write clear instructions in plain English.

Skill files are written in plain English using Markdown format. If you can write clear instructions for a human colleague, you can write a skill.

Business functions where Skills are already being used:

Every major AI platform now supports project folders with persistent instructions: ChatGPT Projects, Claude Projects, and Gemini Gems. Even if you never touch a command line, these tools give you the same core benefit of packaging reusable context.

Checklist: How to Start Building Skills for Your Team

Use this as a step-by-step framework for identifying, writing, and deploying your first set of AI Skills.

Phase 1: Identify What to Build

1. Audit your team's recurring AI prompts

Survey each department. Ask: "What do you ask AI to do at least three times per week?" Start with a task you perform at least three times per week. The repetition gives you enough iterations to refine the skill quickly, and the time savings compound immediately.

Common starting points include:

2. Prioritize by repetition and inconsistency

The best first Skills are tasks where:

3. Document the "best version" of each workflow

Before you write anything for an AI agent, write it for a human. What steps does your best performer follow? What do they check? What does "done right" look like? A good skill encodes knowledge the model does not already have. Large language models already know how to write Python, structure JSON, and compose emails. They do not know your company's deployment checklist, your team's review criteria, or your product's specific quirks.

Phase 2: Write Your First Skill

4. Start small: one task, one file

Your first skill can be 10 lines of Markdown. You do not need scripts, assets, or complex workflows to get value. A simple Skill that defines how to format a client update email will save your team hours in the first week.

5. Nail the description (this is what makes or breaks it)

If your skill does not trigger, it is almost never the instructions. It is the description. The description field is how the AI agent decides whether to load your Skill.

The description is critical for skill selection. The agent uses it to choose the right Skill from potentially 100+ available Skills. Your description must provide enough detail for the agent to know when to select this Skill.

Write it like a routing rule, not a title. Include both what it does and when to use it.

6. Structure instructions as numbered steps

Structure your instructions as clear, numbered steps. Include specific details about what to check, what to output, and how to format results.

Each step should answer: what does the agent do, what does it check, and what does the output look like?

7. Include examples of good output

Show the input pattern and the expected output pattern. Examples reduce ambiguity far more effectively than extra prose. If your Skill generates emails, include one example of a correctly formatted email. If it produces reports, show the expected section structure.

Phase 3: Keep It Lean

8. Stay under 500 lines in the main file

Keep SKILL.md (a markdown file format) under 500 lines. Move detailed content into references, scripts, or asset directories. Remember that the context window is shared space. Every token your skill consumes is a token unavailable for the conversation, other skills, and the agent's reasoning.

9. Use progressive disclosure

Not everything needs to load at once. SKILL.md files serve as an overview that points the agent to detailed materials as needed, like a table of contents in an onboarding guide.

If your Skill covers multiple scenarios (e.g., different report types for different clients), put the specifics in separate reference files. The agent loads them only when needed.

10. Only add context the AI does not already have

Challenge each piece of information you include. AI models already know how to write emails, format markdown, and summarize text. What they do not know is your company's specific formatting requirements, your brand voice, your compliance rules, or your approval workflow. Focus your Skill on that proprietary knowledge.

Phase 4: Test and Refine

11. Run the Skill on real tasks before sharing it

Run the skill on real tasks, refine weak spots, then add it to the team directory so the workflow is discoverable and repeatable. Do not deploy based on one successful test. Run it on 5-10 real examples and look for where it breaks.

12. Track failures and update

The best signal for skill improvements is when the agent picks the wrong skill or executes it incorrectly. Track these failures. For each failure, update the SKILL.md to address the root cause. Skills improve through iteration, not initial design.

13. Validate the Skill with your team's AI tool

You can test a Skill before deploying it by feeding the full SKILL.md to your AI tool and asking it to simulate execution. Feed the model your entire SKILL.md and directory structure, then ask it to simulate step-by-step execution based on a real task and flag any points where it would be forced to guess.

Phase 5: Deploy and Scale

14. Make Skills discoverable

Skills only help if team members know they exist. Build discoverability into your team workflow: maintain a skills list with one-line summaries, include skills usage in onboarding docs, and announce new skills in your team communication channels.

15. Assign a human owner for each Skill

Skills encode your best practices, but someone on your team needs to own and audit what those practices are. When processes change, client requirements shift, or compliance rules update, the corresponding Skills need to be updated too. Treat Skills like living documentation, not "set and forget" templates.

16. Start with 5-10 Skills, then expand based on demand

Most teams find value with 10-30 skills. Beyond that, discoverability becomes a bottleneck. Start with 5-10 skills covering your most-repeated workflows, then add new ones based on actual demand.

Where to Start This Week

You do not need a developer, a command line, or a paid subscription to begin. Here is a 30-minute exercise:

That is your first Skill. It may not live in a formal SKILL.md file yet, but the practice of packaging reusable AI instructions is the same regardless of format.

Projects are the onboarding. Skills are the training. Automations are the job done every day. And agents are the coworker you interact with to get it all done.

- Annie Duke, Thinking in Bets

When the process is right, good outcomes follow more often and more predictably.

Stay in touch with our team from anywhere.