Agent Frank

Managed SaaS $499/monthly

Frank handles 60% of the SDR job — prospecting, writing, sending, follow-up — at roughly 10% of human SDR cost. The other 40% (reply intent, complex objections, meeting prep) still needs a human. Best value in the AI SDR category if you know what you're buying.

OVERALL SCORE

7.0

out of 10

Features 6.5/10

Value 8.5/10

UX 7.0/10

Data quality 7.0/10

TL;DR

Agent Frank is the AI SDR built by Salesforge that handles the mechanical half of outbound sales: it finds prospects, writes the emails, sends them, and follows up — in auto-pilot mode without daily input, or in co-pilot mode with your approval before each send. At $499/month with a 500M-contact database included, it is the best-priced “full” AI SDR on the market.

What it cannot do is equally important: email only, no LinkedIn, no phone, and reply handling that tops out around 75% accuracy on intent classification. Frank is a 0.6 SDR equivalent, not a 1.0.

At a glance

Dimension	Agent Frank
Channels	Email only
Autonomy modes	Auto-pilot (full) · Co-pilot (approval-gated)
Pricing	$499/month (1,000 contacts, ~7,000 emails)
Contact database	500M+ via Leadsforge
Email infrastructure	Not included (Infraforge $33/mo+ or BYO)
Languages	20+
Reply handling	~75% intent accuracy
LinkedIn / phone	✗
Buying signals	✗

What Frank actually does

Frank runs a three-step loop per prospect: research → write → send.

Research. Frank pulls firmographic and contact data from its Leadsforge database (500M+ contacts). It identifies company size, industry, role, and recent context signals where available — enough to drive light personalisation.

Write. The email generation runs a research-then-write loop using Claude under the hood. The output reads like a competent SDR’s third draft: rarely embarrassing, not frequently exceptional. For broad ICPs, the quality-to-volume ratio is strong. For narrow technical ICPs where domain nuance matters, co-pilot mode with manual review before send is the honest setting.

Send and follow up. Frank handles sequence management autonomously: initial email, follow-ups on configured intervals, soft stops on reply or bounce. You set the cadence rules once; Frank executes them without reminders.

This covers prospecting, writing, sending, and follow-up — roughly 60% of a traditional SDR’s job. The other 40% (reading reply intent correctly, handling objections, coordinating handoffs, relationship warmth) still requires a human.

Auto-pilot vs co-pilot

This distinction matters more than most vendors admit. Frank offers both:

Auto-pilot: Frank sends without human review. Maximum throughput, minimum oversight. Appropriate when your ICP is broad, your message variance is low, and deliverability is already dialled in. The risk is that errors in persona targeting or copy quality compound across hundreds of sends before you notice.

Co-pilot: You approve before Frank sends. Adds 5–10 minutes of daily review per 50–100 emails. The honest mode for most teams — catches Frank’s edge-case errors before they become sent emails. Co-pilot is underrated in the AI SDR category, where full autonomy is often oversold.

Start in co-pilot. Graduate to auto-pilot after 2–3 weeks of approval runs show consistent quality.

Email quality

Frank’s email quality is its strongest differentiator. G2 reviewers consistently rate the output above what their human reps wrote — the personalisation depth goes beyond name-and-company tokens, and the sentence rhythm avoids the tell-tale AI cadence most tools produce.

The underlying model (Claude) is visible in the output for trained readers. For buyers, that’s largely a non-issue. For SDR teams using Frank to augment human writing, the quality floor is high enough to ship without heavy editing on broad ICPs.

Technical ICPs are the exception. Frank can identify “platform engineer at a Series B fintech” but cannot reliably distinguish the nuance between a Kubernetes-heavy infrastructure company and a payments-heavy product company. Generic copy reappears at the exact moment specificity would matter.

Reply handling

This is where Frank’s limitation is most consequential. Intent classification runs at approximately 75% accuracy:

Clear positives (“yes, book a time”): caught reliably
Clear unsubscribes (“remove me”): caught reliably
Information requests (“can you send more detail?”): frequently mislabelled as neutral/negative
Ambiguous replies (“interesting, let me think”): unreliable

A mislabelled information request means a warm lead goes cold because Frank treats it as a stop signal. At scale, 25% misclassification rates on reply intent add up to real missed pipeline. Monitor your reply inbox in the first 30 days.

Pricing

Cost	Amount
Agent Frank platform	$499/month (billed monthly) or ~$416/month (annual)
Infraforge email infrastructure	$33/month+ (or BYO domain/SMTP setup)
Leadsforge database overage	Additional, if contact limits exceeded
All-in, typical	$530–$650/month

At $530–$650/month all-in, Frank costs roughly 10× less than a junior SDR fully loaded. The ROI math works if Frank generates at least 1–2 qualified meetings per month — a conservative benchmark for most outbound motions.

When to pick Agent Frank

Email is your primary (or only) outbound channel and you are not actively prospecting on LinkedIn
You have no SDR headcount and need to cover the prospecting + writing + sending motion
Your ICP is broad enough that high-volume email with moderate personalisation works — professional services, SaaS selling to broad verticals, agencies
Budget is $500–$700/month all-in and you want the most autonomous option in that range
You’re willing to handle reply intent manually or accept the 75% auto-classification rate

When not to pick Agent Frank

Your buyers live on LinkedIn — Artisan or manual SDR work is better
Your ICP is narrow and technical — personalisation depth won’t hold
You need multi-channel sequences (phone, SMS, WhatsApp) — look at 11x or human-in-the-loop tools
You want buying-signal triggers (funding rounds, hiring patterns, intent data) built in — Frank doesn’t have them

FAQ

What does Agent Frank actually do end to end? +

Frank finds prospects from its 500M-contact database, writes personalised emails using a research-then-write loop, sends via your email infrastructure, and follows up autonomously. In auto-pilot mode it runs without input; in co-pilot mode you approve before each send. It stops at the booked meeting — post-meeting nurture, complex objections, and closed/won stages are out of scope.
How good is the email quality really? +

Genuinely above average for the category. G2 reviewers consistently report the output reads better than what their human SDRs wrote, and the personalisation goes beyond first-name tokens. The underlying model (Claude) is visible in the sentence structure. At high send volume and broad ICPs, quality holds; at narrow technical ICPs where nuance matters, generic copy reappears.
What does $499/month actually include? +

$499/month covers the Frank platform for 1,000 active contacts and roughly 6,000–7,500 personalised emails per month. Email infrastructure is not included — you need Salesforge's Infraforge ($33/month+) or your own domain/sending setup. Lead database credits via Leadsforge are additional if you exceed included limits.
Is Frank a good fit for technical ICPs (SaaS, fintech, devtools)? +

Partially. Frank's research loop can identify company type and role, but cannot distinguish a Kubernetes platform engineer from a payments platform engineer with enough nuance for truly technical copy. For technical ICPs where the difference matters, co-pilot mode with human prompt review produces better results than pure auto-pilot.
How does Frank compare to hiring an SDR? +

A fully loaded junior SDR costs $60,000–90,000/year in most markets. Frank at $499/month ($6,000/year all-in) handles roughly 60% of that role's output. The 40% gap covers reply intent, objection handling, relationship warmth, and the judgment calls that move mid-funnel deals. Frank is not a 1:1 replacement — it is a force-multiplier for a founder or team without SDR headcount.

AGENT FRANK HEAD-TO-HEAD

Agent Frank review